SOCIETAS LINGUISTICA EUROPAEA

Since the elections of 1986 in Ohrid the following persons nowserve as officers or members of Committees:

PRODUa.

NEW-WOR

President:

Vice-President:

Secretary/Treasurer :

Editors: FoL:

FoLH:

Thomas V. Gamkrelidze, Tbilisi

Pavle lvio, Beograd

Werner Winter, Kiel

Wolfgang U. Dressler, Wien

Jacek Fisiak, Poznan

Executive Oommittee: Maria-Elisabeth Conte, Pavia

Svenka Savio, Novi Sad

Ruta Nagucka, Krak6w

Gilbert Lazard, Paris

Francisco Marcos Marin, Madrid

E. F. Konrad Koerner, Ottawa

Harm Pinkater, Amsterdam

The question of hoy

brew is of interest frorr

interrelated issues of

nature of linguistic

hand, and the distinc1

productivity, or betw'

of inflectional morpho

other. The latter quest

data from noun-comp

vid 1986). Here, our f

in current Hebrew usa

my own and of other

especially Clark & Bel

Modern Hebrew aff

this issue on both extr:

of internal structure.

treme instance of "digl

prescriptive or official

usage manifested by l

tion - as demonstra

Nahir 1978; Ravid in

a survey of this area,

inter alia, in the relat

loni-Feinberg 1974, Ni

in the present study, .

devices that are offici

Hebrew Language Ac

grammars recommend

Publications Oommittee: J6zsef Molnar, Budapest t

Paul Valentin, Paris

Jacek Fisiak, Poznan

Martin Harris, Salford

Irene Philippaki-Warburton, Reading

Wolfgang U. Dressler, Wien

Nominating Oommittee: R. R. K. Hartmann, ExeterGeorg Bossong, Munchen

Martti Nyman, Helsinki

Anna Giacalone Ramat, Pavia

Martin Harris, Salford

Oorrespondence concerning membership and other administrative mattersshould be Bent to: Prof. Werner Winter

Olshausenstr. 40-60

D-2300 Kial

West Germany

Fees and payment modalities:

The annual membership dues are DM 35.00; the admission fee is DM 9.00.

Payment should be made not later than 1 March of the year for whichmembership rights afe desired to the following account:Postal checking account nr. 2428 67.201, Postscheckamt Hamburg (Professorw. Winter, Kiel).

Ifpayment is made in a currency other than German marks, the equivalentofDM 35.00 (or DM 9.00, respectively) should be increased by ten percent tocover conversion charges.

IrBOnS now PRODUCTIVITY IN THE LEXICON:

NEW-WORD FORMATION IN MODERN

HEBREWisi

RUTH A. BERMAN

I. INTRODUCTION

DM 9.00.

The question of how new words are constructed in Modern He­

brew is of interest from several perspectives. The topic bears on two

interrelated issues of concern to current linguistic theory: The

nature of linguistic pro d u c t i vi t Y in general, on the one

hand, and the distinction between s y n t act i c and I e x i c a I

productivity, or between the rules of grammar - including those

of inflectional morphology - and the domain of the lexicon, on the

other. The latter question is the topic of a separate study, based on

data from noun-compounding in Modern Hebrew (Berman & Ra­

vid 1986). Here, our focus is on lexical productivity as manifested

in current Hebrew usage, as an extension of prior, related studies of

my own and of others (Berman 1982; Berman & Sagi 1981; and

especially Clark & Berman 1984).'

Modern Hebrew affords a particularly good case for analysis of

this issue on both extraneous sociolinguistic grounds and for reasons

of internal structure. Thus, the language represents a rather ex­

tremeinstance of "diglossia" as between the puristic requirements of

prescriptive or official norms compared with the actual colloquial

usage manifested by native speakers of different levels of educa­

tion - as demonstrated in field-studies by Donag-Kinrot 1978;

Nahir 1978; R.avid in preparation; and Schwarzwald 1981 (and for

a survey of this area, see Rabin 1984). This disparity is reflected,

inter alia, in the relative acceptability of lexical innovations (AI­loni-Feinberg 1974, Nir 1982). One question which will cOncern us

in the present study, then, is the extent to which word-formationdevices that are officially sanctioned - by the authority of the

Hebrew Language Academy; by usage manuals and prescriptivegrammars recommended for the schools; and by accepted lexico-

rid

'a

�eading

Te matters

for which

(Professor

lquivalent

s>ercent to

0165 -4004/87/0021-425 $ 2,­@ Mouton Puhlishers, The. Hague;

Societas Linguis ica Europaea

�

I,o""'",' .•

426

graphic practice - are reflected in the way speakers in fact con­

strue new words in their langnage.

A second extraneous factor which makes Hebrew of interest in

this connection derives from the peculiar socio-historical circum­

stances attendant on the revival of Hebrew as a spoken vernacular

in the past eighty to a hundred years. There has been an immense

spate of new-word formation activity in the language - including

the early efforts of Eliezer Ben-Yehuda, David Yellin, and other

Hebraists in turn-of-the-century Palestine alongside of more con­

temporary innovations to cover computer, space-age, and technical

terminology in general; current political, economic, cultural, and

other media-oriented coinages; as well as slang and similar in-group

usages developed among schoolchildren, soldiers, students, and

the like_

Finally, in more strictly structural terms., the nature of word­

formation is of interest in view ofthe kind offormal devices available

to Hebrew speakers for this purpose. Thus, Hebrew is a relatively

synthetic, rather than agglutinating or isolating langnage, with a

complex system of bound morphology and a rich array of lexica­

lization devices, including the peculiarly Semitic method for form­

ing new words by means of consonantal roots associated with a

large, but finite, set of affixal pattsrns_ The main devices serving

this function can be ranked in terms of the relative degroo of

incorporation or separation of different morpho-syntactic constit­

uents, as follows:

(I) Illustration of Major Devices for New-Word Formation in

Hebrew'

L Zero-Affixation:

lConversion]

2. Fused Affixation:

[Root + Pattern]

3. External Affixation:

[Stem + Affix]

4_ Blending:

[Stem + Stem]

5. Compounding:

[Word + Word]

menaheZ V= (he) directs, N=director

bolet V=protrude, Adj=conspicuous

nahal CoCaO = procedure

bUta CCiCa = protuberance

nohal-iy = procedural

bolt-ut = salience

zrak-or = throwlite = projector

rakevel = traincable = cable-car

zorek diskua = discus-thrower

rakevet taxtit= train-under= subway

2.

Against this back�

Hebrew speakers co:

and potential lexico

innovative and con�

native speakers of H.

subjects aged 20-3

guage, and 10 eleve

preciable differencei'

so their results are tl

The tes-t conCemCi(

like the real words m

'student' (soo n. 2

misparayim 'scissOr!

chine = typewriter';

rant' maZon 'hotel',

nouns - e.g. takZit

kaduregeZ 'football te

'purity', aclut 'lazin.

with three different j

(2) A = Production

Subjects w<

in Hebrew

always hu!

you calla Cl

B - Selection 0

Subjects WE i

10 each as :does a cartl

and a locatl,

C - Listing of (I

Subjects we \came to thE.

noted abov.

I

speakers in fact con-

Hebrew of interest in

wio-historical circum­

s a spoken vernacular

has been an immense

language - including

vid Yellin, and other

ongside of more con­

Ice-age, and technical

momie, cultural, and

: and similar in-group

diers, students, and

the nature of word­

mal devices available

ebrew is a relatively

ing language, with a

rich array of lexica­

tic method for form­

Os associated with a

nain devices serving

l relative degree of

lO-syntactic constit-

vord Formation in

directa, N=director

" Adj=conspicuous

rocedure

ltuberance

lural

te = projector

ble = cable-car

lCUS-thrower

,in-under = subway

.. ",

2. DESCRIPTION OF STUDY

Against this background, we undertook to investigate how native

Hebrew speakers construe different classes of nouns in the actual

and potential lexicon of their language. We constructed a test of

innovative and conventional nouns, administered in writing to 28

native speakers of Hebrew - 18 college students or college-educated

subjects aged 20-30, non-experts in linguistiCS and Hebrew lan­

guage, and 10 eleventh-graders aged 17-18." There were no ap­

preciable differences between the responses of these two groups,

so their results are treated together below.

The test concerned five classes of nouns: (i) Age n t nouns ­

like the real words menahel 'boss, manager', xacran 'janitor', 8t'llAUnt

'student' (see n. 2 above); (i1) Ins t rum e n t nouns - e.g.

misparayim 'scissors', iparon 'pencil', Jn3XJnat-lctiva 'writing-ma­

chine = typewriter'; (iii) Pia c e nouns - like mi8'ada 'restau­

rant' maZon 'hotel', xadar-8heyna 'bedroom'; (iv) Collective

nouns - e.g. taklitiya 'record-collection', gedud 'troop', kvucat­

kaduregel 'football team'; and (v) A b s t r act nouns - e.g. t6har

'purity', aclut 'laziness', cima'on 'thirst'. Subjects were presented

with three different tasks, presented in the following order:

(2) A = Production of innovative coinages - 40 items

Subjects were given definitions of words that do not exist

in Hebrew - e.g. "What would you call a person that's

always hugging 1" = ha-marbe le-xabek; "What would

you call a collection of balloons!" = 08ef 8hel balonim.

B - Selection of innovative iteins - 30 items

Subjects were asked to select 30 out of a total of 52 items,

10 each as best suited to being names of; a person tbat

does a certain job; an instrument, utensil, or machine;

and a location, respectively.

C - Listing of conventional words 35 items

Subjects were asked to write down the first 5 or 10 words

came to their minds for each of the five classes of nouns

noted above.

428

The items used lIB the bllBis for new coinages in Part A were taken

from a list of 40 commOn verbs used in a prior study of how children

and adults comprehend and produce innovative agent and instru­

ment nouns in Hebrew (Berman, Hecht & Clark 1982), while the

forms provided in Part B were based on findings of thiA study com­

bined with the devices typically lIBsociated with the various

clllBses of Hebrew nouns, as further specified below. In general, the

questionnaire WllB constructed to test a series of hypotheses about

the lexicon in Modem Hebrew, deriving from the following sources:

� Examination of entries for th""e different clllBses of nouns in

sources concerned primarily with more normative written usage,

including a major standard monolingual Hebrew dictionary (Even­

Shoshan 1979); a Hebrew-English dictionary bllBed on frequency

counts (Balgur & Dagut 1975); listings of Hebrew noun patterns

(Avinery 1976, Barkali 1964, Rabinowitz 1947); and studies of spe­

cific clllBses of nouns (Du Nour 1979, Gluska 1981, Oman 1979);

- Findings of studies which examine different lIBPects of the

current Hebrew lexicon: Alloni-Feinberg 1974, AttillB 1980, Berman

& Ravid 1985, Bolozky 1978, Donag-Kinrot 1978; Nir 1982; Ravid

1978; Werner 1982, 1983;

- Results of small-scale studies of how speakers coin new terms

for specific subclllBBeB of nouns conducted by students of mine in

the context of class projects in lexicology; and

- Results of a prior study of 60 children aged three to twelve

years and of 12 adults on an oral tlIBk requiring them to construct

and to interprete innovative agent and instrument nouns in Hebrew

(Clark & Berman 1984).

3. HYPOTHESES AND FINDINGS

Below we present the main findings of the questionnaire outlined

in (2) above, from the point of view of: the options preferred for

coining new terms (Section 3.1); how these accord with normative

dictates (3.2); the role of the conventional or well-established lexicon

(3.3); types of structural devices favored by speakers (3.4); the

status of compounding as a word-formation device (3.5); and the

relative transparency or distinctiveness of the different noun

classes we examined (3.6).

3.1 Preferred Options

We assumed that Sp<

formation in their Ian!

acroSS a random collel

fair amount of a g r ,

ent tlIBks. On the oth,

diverge considerably f

and that this discrep"

were required to inno.

(Part A of the test) tl

most suitable forms (F

out by the major respe

naire, lIB set out in Tal

Coinages for the six

are not included in T,

category. Over 10% 0

lIB compared with onl�

Most highly favored res

(Part A), given in percel

Device �i

CaCCan

Stem-an

Verb-an

Final-an

m-=Verb

roscCec

miCCaC

mijmaCCeCajet

... iya/yada

xabka1i

sta1kan

mistakJ

mekac< I....aId

mialak Ima'a:tf.

mular. !!

sadra71 J

Takavi �

(I) Note: Horizontal

class of noUllS. Figures

es in Part A were taken

If study ofhow children

.tive agent and instru­

Clark 1982), while the

lings of this study com­

ted with the various

t below. In general, the

es of hypotheses about

I the following sources:

'ent classes of nouns in

mative written usage,

.rew dictionary (Even­

y based on frequency

Iebrew noun patterns

7): and studies of Bpe­

1981, Oman 1979);

fferent aspects of the

" Attias 1980, Berman

1978; Nir 1982; Ravid

lakers coin new terms

7 students of mine in

d

aged three to twelve

ng them to construct

lent nouns in Hebrew

NGS

uestionnaire outlined

)ptions preferred for

cord with normative

II-established lexicon

speakers (3.4); the

[evice (3.5); and the

the different nOun

429

3.1 Preferred Opti.0n8

We assumed that speakers' preferences with respect to new-word

formation in their language would not be haphazardly distributed

across a random collection of devices, and that there would be a

fair amount of a g r e erne n t among subjects across the differ­

ent tasks. On the other hand, we felt that speakers' choices might

diverge considerably from official norms for new-word formation,

and that this discrep'lJlcy would be more apparent when speakers

were required to innovate freely by producing forms of their own

(Part A of the test) than when they were asked to select or judge

most suitable forms (Part B). These assumptions were largely borne

out by the major response-pattsrns on the first part of the question­

naire, as set out in Table I below.

Coinages for the six items intended to yield A b s t rae t nouns

are not included in Table 1, since this proved to be a problematic

category. Over 10% of the items received "no response" blanks _

as compared with only 4% no responses across the other categories;

Table 1

Most highly favored response types in the production of innovative forms(Part A), given in percentages for four different claas88 of nOUDS [N = 28]

Noun 01a8s

IleTioo Sampl.

I�I I-""""Aoen'"

"""'"Colleo

.. Iterr.ul: 28' m '88

CaeCan xabkan hugger 48.2

Stem-an atalkan fleer 7.9

Verb-an m",laklan looker 27.5

Final-an.............. 83.6(1)

m-=Verb mekac'c cutter 7.1maCCeC madl,. lighter

21.41miCCaC mialak run-place 8.6

mi/maCCeCa/et ma.'axela eating.utensil 11.9

26.4 32.9

mixla11Ul dream-place --- ---

54.9 41.5--- ---

... iya/yada sadraniga arranging-place

rakaviydda train-collection 30.8 57.7

--- ---

(I) Note: Horizontal lines mark off the most favored responses for eachclass of noWUJ. Figures entered between lines are sums of several subclasses.

430

and subjects interpreted these as a d j e c t i v e s in over 20%

(35 out of 168) of their responaee - giving either real words, e.g.

matok 'sweet' for the quality of something 'taaty' (Hebrew ta'im),

margia 'soothing' as the quality of something 'blue' (kaxol),

meyushan 'antiquated' for the quality of a thing that is old (yashan

_ with the same root), or else novel adjectives such as yashin

'old-y', or participial-like forma - e.g. menmnax 'shorted' for the

quality of a person who is short (namux). In general, results on

this subset yield a very mixed picture, as follows. Around 40% of

the coinages took one of three forma commonly used for abstract

nouns in Hebrew, thus: 22% were given the suffix out, as in innova­

tive te'im-ut 'tasti-ness', raz-ut 'thin-ness'; another 10% (17 words)

got the ending -on as an external suffix (e.g. nemux-on 'short-ness')

or as part of an affixal pattern, e.g. kixalon 'blue-ness', as in con­

ventionalshiga'on 'madness', (but 11 of these were non-innovative

- the single item rawn, which is the conventional word for 'thin­

ness'); and another 12 words (7%) took a vowel-internal C6CVC ­

e.g_ innovative nomex 'shorty-ness', ta'am 'taste-ness', r6zi 'thinny.

ness'. Other answers included items ending in -an - mostly (16 out

of 24) for the single item defined with a verb rather than an adjec­

tive to name the quality ofa person who is constantly falling (notel),

as well as numerous more idiosyncratic forms, depending on the

particular input item in each case. It thus seems that this noun

class - at any rate as represented in the task at hand - did not

evoke any single response or class of responses as most favored for

new-word formation. In the subset of "abstract" nouns, rather,

inter-subject agreement was manifested in the high number of

identical answers given to a specific item - as noted for conven­

tional razon 'thinness' and innovative natlan 'faller/falling' above.

:elsewhere, our hypothesis of "agreement" is largely confirmed

by how subjects performed when deriving innovative nouns from

verbs. This is overwhelmingly the case for the 10 Age n t noun

items randomly distributed across the 40 definitions constituting

Part A: Over 80% of these coinages ended in the syllable -an ­

exactly corresponding to the clear preference for this ending shown

in innovative agent nouns in our earlier, oral study (Clark & Ber­

man 1984). Such forms also account for nearly a quarter of all the

Instrument nouns coined as well, although these yielded a more

varied picture than the agents, as follows: Around one-quarter were

forms ending in -an; another quarter were masculine nouns with a

prefixal m-; and a q'

nine suffix (stressed

variety of forms typi

discuBBed below. P I

feminine nouns with

ventional option of •

other non-native fen

half of all the innova

Beyond these mail

class manifested at Ie

Agents - 5.5% be

Instruments - 11.5

(e.g. mpar-it for a roo

- 10% forms with

responses varied, COl

coined for the six c

- 4.5%; real word,

compounds - 3.5%.

Clearly, then, resp'

asked to coin name.

across the five clas,;

and abstract states' .

but showed a coneid.

the subjects. Moree,

accord only partially i

innovators (Section

(3.3 below).

3.2 Comparison with

Our next hypoth.. i

fully in accord with t !fOrIDS for coining ne ICaCaO for agent-ocOl I

• I,

ner'; maCCeC for In I

'amplifier'; miCeaC,

mirpa'a 'clinic'; and C

caneret 'pipeworks'.

test, where subjects

vative fOrIDS with ...

tives in over 20%

either real words, e.g.

tasty' (Hebrew ta'im),

ething 'blue' (lcaxol),

ing that is old (yashan

lCtives such as yashin

Umaz 'shorted' for the

In general, results on

,llows. Around 40% of

only used for abstract

:mffix -ut, as in innova­

lOther 10% (17 words)

nemux-on lehort-ness')

'blue-neas' J as in Con­

e were non-innovative

ltional word for 'thin­

weI-internal C6CVC ­

ate-ness', rozi 'thinny­

I -an - mostly (16 out

, rather than an adjec­

IStantly falling (notel),

ms, depending on the

seems that this noun

,k at hand - did not

es as most favored for

ltract" nouns, rather,

the high number of

as noted for conven­

, 'faller/falling' above.

, is largely confirmed

lDovative nOUDS from

he 10 Age n t noun

,finitions constituting

in the syllable -an ­

for this ending shown

I study (Clark & Ber­

ly a quarter of ,,11 the

these yielded a more

lUnd one-quarter were

asculine nouns with a

prefixal m-; and a quarter were words with prefixal m- and a femi­

nine suffix (stressed -a or unstressed -et). This reflects the greater

variety of forms typical of instrument nouns in general - as further

discuesed below. P I ace nouns also selected as high as one-third

feminine nouns with prefixal m-; another third took the less con­

ventional option of suffixal -iya - an ending which together with

other non-native feminine endings such as -iydda accounts for over

half of all the innovative Colle c t i v e nouns as well.

Beyond these main trends, as shown in Table 1 above, each noun

class manifested at least one other relatively favored response, thus:

Agents - 5.5% benani (present-tense, participial) verb forms;

Instruments - 11.5% various suffixes including -iya and also -it

(e.g. xapar-it for a machine used for digging = la-xpor); Collectives

- 10% forms with plural endings or other suffixes, etc. Other

responses varied, coming to around only 10% of the Part A forms

coined for the six different noun classes, as follows: No answer

� 4.5%; real words (I.e. failure to innovate) - 4%; blends and

compounds - 3.5%.

Clearly, then, responses given by a large group of native-speakers

asked to coin names for a variety of items, randomly distributed

across the five classes of agents, instruments, places, collectives,

and abstract states were by no means unmotivated or haphazard,

but showed a considerable degree of clustering or agreement among

the subjects. Moreover, as we note further below, these responses

accord only partially with the specifications of official or normative

innovators (Section 3.2) as well as of the conventional lexicon

(3.3 below).

3.20omparison with Normative Requirements

Our next hypothesis was that innovative coinages would not be

fully in accord with the prescriptive requirements as to the "desired"

forms for coining new terms in each lexical class - for instance,

CaCaO for agent-occupations - e.g. pasal 'sculptor', ganan 'garde­

ner'; maCCeC for instrument nouns - e.g. mazleg 'fork', mayber

'amplifier'; miCCaCa for place-nouns - e.g. mis'ada 'restaurant',

mirpa'a 'clinic'; and caeeCet for collectives - e.g. tayeset 'squadron'

caneret 'pipeworks'. We assumed that responses in Part B of the

test, where subjects were required to judge the suitability of inno­

vative forma with respect to nouns in the different classes rather

432

than to coin new items themselves aa in Part A" wonld be' a truer

reflection of selfconscions norms for how words, "should" be con­

structed. And this, in fact, proved to be the case, as shown by the

breakdown of results for Part B of the test, where subjects were

required to select for each of the three classes - Agent, Instrument,

Place - 10 items out of the innovative forms presented to them

(30 out of a total 52).

Table 2

Distribution of forms selected as innovative itemB suited to three classes

of nouns (Part B), given in percentages for five morphological categories

[N = 28]

Morphological Ca.tegorifl8O

� I ma. ..(oj I ml..(a) J I,ya I cvovC I V"" i Total-

Sample forma d""... .........m._

.......,..

-"'" "'-1d......., en.mpl-er catch-et fa...."

.- =:tl-�. -....... mimltl1d tmllniJG

""""ad........, ,........, ........ pietme-y h""""" """'''''1Agent 49% 1.3% - I - 19.7% 19.2% I 89.2%Instr 11% 42.5% 10% + 3% 16.0% 16.0% 98.4%

Place 1.8% 28.6% 40%

I16.7% 5.7% 1.0% i 94.1%

I , I

Note: *Each of these I) morphological categories was represented by10 items. Percentages in the table were calculated out of the total

number of items selected for that category. Thus, if each of the

28 subjects selected 10 a+tent nouns, then the total responses =280 for agent nouns.

*. The fact that the totals donot add up to 100% is due to occasion­

a.l selection of other forms presented on this part of the test,

outside of the six options listed here - e.g. kosh1-ut 'failing-ness',

pikaxon 'clevel'-l;y'.

The findings for Part B show a clear trend to differentiation

between the three noun classes: Over 40% of all responses selected

words ending in -an for Agents, words beginning with ma- for

Instruments, and words beginning with mi- for Places. On this

task, however, speakers selected a variety of innovative forms well

beyond the range of those which they deployed in creating coinages

of their own in Part A. Specifically, in the previous task, subjects

had conspicuously avoided options which are less "transparent",

in the sense of manifesting overt one-to-one relations between a

given lexical class of nann and a given stem-external affix to denote

that class (and see further Section 3.4 below). Yet here, in. Part B,

many of the option

ment nouns were I

plus vowel alternat

those with a benan:

tense verb. Thus, "

for as few as 7% of

and instrument nc

30%, or almost on�

respectively). More

proportion of the I

sponding maCCeC,

required midraaa "

as further discUBBC(

The distinction 1

A and the more "n

Part B is yielded l

on these two tasks,

Distribution

innovations in opE

forms present

01...

Agents

InstMunents

Places

These findings an

the normatively pre

established by caref

such as those listec

that the official coi

Parl A, would be a truer

JV worda. "should" be con­

the caae, ll8 shown by the

test, where subjects were

3Ses - Agent, Instrument,

forms presentsd to them

items suited to t.hree classesive morphological categories

ztUhid

8U8pectee-lot", Icatalog.16t I

nuu1U;i1 It1".ad�

19.2% r- 89.2%'16.0% I 98.4%1.0% 94.1%

m=b_�

19.7%

16.0%

5.7%

gories was represented by� calculated out of the totalegory. Thus, if each of the�hen the total responses =

:t to 100% is due to occasion.on this part of the test.

- e.g. koshl-ut 'failing-ness',

orend to differentiation

of all responses selectsd

"'ginning with ma- for

.i- for Places. On this

f innovative forms well

red in creating coinages

previous tll8k, subjects

tre Jess "transparentJ',

Ie relations between a

'xtsrnal affix to denote

l Yet here, in Part B,

many of the options which were selected for both Agent and Instru­

ment nouns were relatively more ops.que forms: those where root

plus vowel alternations yield interdigital CVCVC surface fOrnIB and

those with a benoni participial form which could also be a present­

tense verb. Thus, wherell8 tog e the r these two devices account

for ll8 few ll8 7% of all responses on Part A (3.5% and 4% for agent

and instrument nouns respectively), they were selected in some

30%, or almost one-third, of the Cll8es in Part B (15.5% and 14%

respectively). Moreover, in Part B, Place nouns showed a higher

proportion of the normative miCCaCa pattern than of the corre­

sponding maCCeCa forms characteristic of colloquial usa!'e (cf.

required midr=a 'sidewalk' alongside of commonplace madrexa ­

ll8 further discussed in Section 4 below).

The distinction between "colloquial" or freer coinages in Part

A and the more "normative" or selfconscious judgements made in

Part B is yielded by comparing the most favored responses given

on these two tasks, ll8 in Table 3 below.

Table 3

Distribution of responses on most favored cate�ories forinnova.tions in open-ended production (Pa.rt A) and m selection of

forms presented (Part B), given in percent&ge� [N = 28]

Cl� I Fonn IFut A

IFut B

(produotlon) (selection)

,

Agents I CaCCan

I48 I 30

Word/Stem-an 35

I20

Others: 17 50,

Instruments I maCCeC 19 I 32

maCCeCa. 16 11..• aD 25

�I11

Others: 46

Places I miCCaqa,e') 21 I 45

maCCeqa) 20

I28

... iya 25 17

Others: 34 10

These findings are not consistent with what we had evaluated ll8

the normatively preferred devices for the different noun c1&BBes, ll8

established by careful examinations of word-lists and other sources

such ll8 those listsd at the end of Section 2 above. We &BBumed

that the official coinages would favor the following breakdown of

434

forms: Age n t s would be evenly distributed between (i) CaCCan

- e.g. batlan 'idler', parahan 'commentator'; (ii) CaevC - e.g.

cayad 'hunter', katav 'reporter' and also kacin 'officer', pakW.

'clerk'; and (ill) Conversion - e.g. aha/u 'judges / a judge' judge',

me'amen 'coaches / a coach'; contrastingly, Ins t rum e n t s

would opt for (i) the maCCeC pattern, as in established mazleg

'fork' , masrek 'comb' and more recent makren 'projector'. macber

'battery'; (ii) some feminine maCCeCa nouns, as in recent mamtera

'sprinkler', mavxena 'test-tube'; and (iii) the so-caHed segolate pat­

tern, as in established deyel 'flag', reaen 'bridle', and newer belem

'brake', heddc 'trigger'; while in normative terms, P I ace nouns

would require either miCCaCa as in mia'ada 'restaurant', mixlala

'coHege', lIdld,less commonly, masculine miccaC nouns like miarad

'office', mitbax 'kitchen', or else they would take compound forms,

particularly with the superordinate head noun bu-, as the bound

form of 'house = place-of' - as in establishedbu-kniau 'synagogue'

bet-xaroaMt 'factory'.

These predictions were not borne out at aH in the open-ended

production task of Part A - even although the questionnaire was

administered in writing, and we had assumed that this medium

would yield more selfconscious renderings than a comparable oral

task had earlier shown to be the case (Clark & Berman 1984). More

surprisingly, these normative options were by no means the only

ones selected in the judgement task given in Part B, either. This

accords weH with findings for the adults who participated in our

earlier, oral study. They had consistently avoided evevc forms for

agents and ma- prefix forms for instruments in a production task,

but when subsequently confronted with such innovative forms in a

comprehension task, they responded by revealing their awareness

of more official norms. Thus, after they had been presented with

several coinages such as maxper 'digger' madlele 'lighter', 1IUUlhbera

'breaker' to indicate instruments, respondents often said things

like "Oh, I should have given that before, too I", or "Oh, that is

the (right/correct/good) way we should talk about instruments".

Such comments, like the discrepancies we found between coinages

(Part A) and selections (Part B) in the present, written tasks, in­

dicate that what speakers do in themselves making up new words

is by no means identical to the set of normatively approved or of­

ficial options available to them at a more selfconscious level of

performance.

3.3 Role 0/ the Com

Here we refer to .

lexicon of users (w

Berman & Ravid 1

mal factors would '

occurrences in actu

erS will be attenti'

ries - where this I

but in relation to t.

referents, on the OJ

minence of the lexi

tried to evaluate tl

came to mind for

Results for the me

Most favored rcl&8S(

Banking I AgeD":N - 280

--

I,t I C.C,C

[37%J

2nd I Benoni

Verbs

[20%1

3rd \ ... a.n/ay

[16%/14% .

---

Others I 13%

When asked to 1 'i

gories, speakers die i

device or morpholc i

down words that .. iI

- for instance, wo I

for a person with a i

nouns like: more 'f, I

er', indicating, fin \form dictated thei

!

mted between (i) CaCCan

ttor'; (ii) CRevc - e.g.

10 kacin 'officer', pakid

'judges I a judge' judge',

19ly, Instruments

as in established mazleg

�kren 'projector', macber

IDS, as in recent mamtera

ie so-called segolate pat­

bridle', and newer belem

, terms, P I ace nouns

ula 'restaurant', mixlala

iCCaC nouns like misrad

d take compound forms,

noun bet-, as the bound

Led bet-beset 'synagogue'

,t all in the open-ended

'h the questionnaire was

Imed that this medium

than a comparable oral

< & Berman 1984). More

, by no means the only

in Part B, either. This

¥ho participated in our

'oided evevc forms for

t8 in a production task,

h innovative forms in a

vealing their awareness

td beM presented with

<dlek 'lighter', mashbera

ents often said things

too !H, or "Oh, that is

lk about instruments".

ound between coinages

sent, written tasks, in­

making up new words

>tively approved or of-

, selfconscious level of

3.3 Role of the Conventional Lexicon It

I1

Here we refer to the status of well-established items in the shared

lexicon of users (what Aronoff 1976 terms "old" words; see, too,

Berman & Ravid 1986). We hypothesized that in this respect for­

mal factors would carry less weight than amount and centrality of

occurrences in actual usage. That is, in their own wordstock, speak­

ers will be attentive to pro tot Y pic a I instances of catego­

ries - where this notion is characterizable not in structural terms,

but in relation to the everyday familiarity or pragmatic salience of

referents, on the one hand, and the accessibility or linguistic pro­

minence of the lexical items which encode them, on the othe•. We

tried to evaluate this by asking subjects to list the first words that

came to mind for each of the classes in question here (Part C).

Results for the mcst favored response-types are given in Table 4.

Table 4

Most favored response types in listing of real words for five

cla.ssea of nouns (Part e), in percentages

1st

Noun claas

g

Agents

I I """"'" ICollectives

IAbstracts

N=280 '80 '40 '40 '40

CaCaC m.CCeC(.} m./miCCec'l ... iy. Adjectives

[37%] [28.5%J [41.5%] [30%J [32%]

Benoni Favorite ... iya Sporadic Action-

Verbs Items Items Nominal

[20%] [25.5%J [20%] [30%] [30%]

... antsy Sporadic Compounds No ... ut

Items Response

I [16%/14%J [21.5% [14%J [30%] [17%]

113°! 24.5% 23.5%I

10% 21%I ,0 I

Rankin

2nd

3rd

Others

When asked to list words known to them in the different cate­

gories, speakers did not make reference to any particular structural

device or morphologica] pattern to start with. They initially wrote

down words that seemed to them "beat exemplars" of a given class

- for instance, words like nagar 'carpenter, zashmelay 'electrician'

for a person with a certain job. Very popular among the agents were

nouns like: mare 'teacher', mehandes 'engineer', masger 'metalwork­

er', indicating, firstly, that semantics rather than morphological

form dictated their choices and, secondly, that there was a high

436

enough level of agreement among these particular items to suggest

a clearly shared type of response-pattern for different subjects.

Interestingly enough, over one-third of the age n t nouns listed

were in the CaCaC pattern. e.g. tabax 'cook', tayas 'pilot' nagar

'carpenter', in marked contrast to what we had found on the two

innovative tasks. This suggests that such nouns are noteworthy as

part of the eatablished rather than of the potential lexicon of agent­

nOUllS in current Hebrew usage. Yet in this matter, too, semantics

waa uppermost: If a subject gave a noun such aa, say, xayat'tailor'

chances are his or her next word on the list would be toteret 'dress­

maker', just as after giving as an instrument noun the word iparon

'pencil', very commonly indeed the subject would then list et 'pen'

and/or sargel 'ruler'. '1'he impact of semantic prototypicality was

even clearer in the listing of ins t rum e n t nouns. Here, the

two most popular items (given by more than 20 out of the 28

subjects I) were the worda patish 'hammer' and kat 'spoon', followed

by some more incidental or sporadic items given by five or fewer

of the subjects - e.g. mexonit 'car', mekarer 'fridge', as well as loan

worda like blender, mikser.

The specific P I ace nouns selected from the conventional vo­

cabulary were, however, more in accord with the coinages of the

earlier sections of the teat, around 40% of the worda given being in

the form ofm...a - e.g. maxbeaa 'laundry', mizbala 'garbage-dump'.

But this set also included relatively many compounds -e.g.

bet-se}er 'house-book' = 'school', ulam rik'Udim 'dance hall', migral

s]JO'rl 'sportafield'.

In contraat to these three sets of nouns - worda for agents,

instruments, and places - speakers seemed to have a hard time

accessing col I e c t i ve nouna in their vocabulary. This may be

because they are not even aware that words like lwutsa 'group',

arema 'heap', or kovec 'set' are in fact members of this particular

semantic category. Thus, several subjects gave the word Hta

'claas(room)' aa the name of a place, but none considered that the

same word could also specify a collection (of people).

The results on Part C strongly confirm our hypothesis that the

devices which speakers favor for innovating new worda are only in

part a reflection ofthe well-eatablished vocabulary. Thus, new agent

nouns today rarely take the surface shape of CaCiC, although Bibli­

cal Hebrew had several such nouns - e.g. kacin 'officer' nasix

'prince' (poasibly aa many as its CaCaC nouns like tabax 'cook', sabal

'porter');'and be

with the same sur

exclusively for C(

e.g. shavir 'break

(and sOO further S

pattern is current:

irrespective of ho'

be available in tl

present-tense verI

very common inde

and far less so £,

yet it is seldom a<:

people-agents. Th.

(exceptions being

_ yet it accounl.

nOuns innovated

proportion of sud

given over half th

to name instrume

Berman 1984). Th,

current research c(

"old" and "new"

established in the

speakers, on the 01

extending this rep<

sion in Section 4 b

3.4 Favored Struct.

We had hypathl

Modern Hebrew w'lalso conventional � 'l

which we had a pri( \favored to least fl,

2) Root-incorporat< I'ing - as illustrate<

This set of hypo I

rent Hebrew exter \device than in earl"natural", aggluth

artlCular items to suggest

n for different subjects.he age n t nouns listedook', tayas 'pilot' nagar

ve had found on the two

nouns are noteworthy as

otentiallexicon of agent­

s matter, too, senlantic8

lOh as, say, xayat'tailor'

t would be toteret 'dress­

lt noun the word iparon

, would then list et 'pen'

tic prototypicality waa

e n t nOUDS. Here, the

than 20 out of the 28

nd leaf 'spoon', followed

given by five or fewer

'fridge', aa well as loan

u the conventional vo­

th the coinages of the

e words given being in

izbala 'garbage-dump'.

y compounds -e.g.

'" 'dance hall', migras

- words for agents,

to have a hard time

abulary. This may be

l like leVU/sa 'group',

ers of this particular

�ave the word leita

, considered that the

,ople).

hypothesis that the

'w words are only in

ary. Thus, new agent

tCiC, although Bibli­

cacin 'officer' nasix

:e tabax 'cook', sabal

�,

-

'porter');' and both Biblical and Mishnaic Hebrew had adjectives

with the same surface form � a pattern which today is used almost

exclusively for coining adjectives with a passive '-able' sense _

e.g. shavir 'break-able = fragile', daliq 'burn-able = flammable'

(and see further Section 4 below). From our point of view, then,this

pattern is currently "open" or lexically productive for this meaning,

irrespective of how many such form-meaning links may happen to

be available in the well-established lexicon. Relatedly, the benoni

present-tense verb form - the option of conversion, that is _ is

very common indeed for well-established, conventional agent nouns

and far less so for instrument nouns (Berman 1978, 394-405);

yet it is seldom adopted as a means for "spontaneous" coinag". for

people-agents. The -an ending is rare for occurrent instrument nouns

(exceptions being potzan '[can] opener', mazgan '[air] conditioner')

- yet it accounted for some one-quarter of all the instrument

nouns innovated in the current study, and for an even higher

proportion of such nouns in our earlier, oral study - where it was

given over half the time by the eleven-year olds and adults asked

to name instruments used to carry out certain activities (Clark &

Berman 1984). These findings clearly support the claims made by

current research concerning the need to distinguish clearly between

"old" and "new" words, hence between items which have become

established in the conventional wordstock of a language and its

speakers, on the one hand, and the currently productive devices for

extending this repertoire, on the other (and see, further, the discus­sion in Section 4 below).

3.4 Favored Structural Devices

We had hypothesized that in a. Yery general way speakers ofModern Hebrew would select innovative - and to a lesser extentalso conventional - items in terms of certain struetural preferences,which we had a priori ranked in the following order, from most highlyfavored to least favored devices: 1) Stem/word + external affix;2) Root-incorporated affix, and 3) Analytic compounding and blend­ing - as illustrated in Section 1 above.

This set of hypotheses was based on the assumption that in cur­rent Hebrew external affixes are taking over as a more productive

device than in earlier times - providing the language with a more"natural", agglutinating kind of option (Dressler 1981). Here, by

438

"productive" we refer to the availability of new structural options

which were not in general use at earlier stages of the language.

These include: the widespread contemporary use of suffixal -iy to

derive denominal adjectives, far beyond its more restricted range

of application in Medieval Hebrew (e.g. recent memshalt-iy 'go­

vernment-al', zorp-iy 'winter-y'); the extension of suffixal out to

express a wide variety of abstract state nouns - as in recent

me'urav-ut <involve-mant' J metuxkam-ut 'sophisticated-ness', xesk­

bona'ut 'accountan-cy'; the addition of -an as an agent marker, not

only on full nouns 8.B in, say, mizrax-an 'orient-aHet', tavru'-an

sanitation-ist', but also with present-tense stems to yield words

like juvenile marbic-an.�hit-ter'. and mera/et-an 'pryer', mistakl-an

'starer'; the extension of the suffixes -iya and the foreign -iytida

for collective and place-names - e.g. well established nagar-iya

'carpentry-shop', merkaz-iya 'central-exchange', and less conven­

tional glida-riya 'icecream-ery', tremp-iytida 'hitchhike-station';

wide use of -on to indicate periodicals - e.g. well-established

shnat-on 'annual', newer mkom-on 'local (paper)' - as well as

diminutives - e.g. xadr-on 'little-room', dub-on .teddy-bear'; and

the extension of the suffix -it not only in forming diminutives like

kos-it 'little-glass', map-it 'napkin = little cloth', but also for a

wide range of food brandnames such as shum-it 'garlic-cheese', and'

laxm-it 'wheat-cracker', or loan-based names for the soft-drinks

trop-it, shoko-lit (Attias 1980). Alongside of all these, a further note­

worthy innovation in stem-external affixation - one not addressed

in our present study - is the current use of prefixes based on

Graeco-Latin loan translations, such as ben-Ie'umi 'inter-national'

rav-goni 'varie-gated', tat-karka'i 'sub-terranean', tlat-memadi 'three­

dimensional' - a device totally foreign to earlier stages of Hebrew.

Despite the extensiveness of such devices, results of our study

reveal that speakers still rely heavily on the classic Semitic device

of consonantal root extraction plus affixation by means of an ac­

cepted morphological pattern of the kind traditionally termed

mishkal. (Implications of this situation for the theory of Natural

Morphology are discussed in Werner 1982.) This was particularly

true for the three noun classes we chose to focus on - Agents,

Instruments, and Places respectively. This is revealed by the

responses to Part A, where subjects were asked to innovate on the

basis of definitions containing verbs in the infinitive form, that is,

with a prefixall- marker, often with an additional stem-prefix as

well (compare,

marteva 'wetter'

below, overwhel

ical affixation ­

affixal pattern .

around one-thin

which we had li

(3) Distribution

!Root + Pattern I:

Stem + Suffix i

IIi

Device

[MORt favored

suffix]

Other devices,

sible responses,

(conventional Ie, !

compounds and I !

than on conversi,

in Part B, where

subjects. Yet eve

tern affixation fa

affix options for

included in this :

listing of famili,

respect. Yet here

basis for many 01

bulary, applying

third of Instrum.

lective and Abet

inconclusive pict,

These findings I

Or "old" wordsb

coinages, Hebre"

morphological, 01

)f new structural options

stages of the language.

ry use of suffixal -iy to

" more restricted range

recent memshalt-iy 'go­

naion of suffixal out to

nouns - as in recent

'phisticated-ness', xeah­

as an agent marker, not

'orient-alist', tavru'-an

, stems to yield words

Jt-an 'pryer', miBtakl-an

and the foreign -iyOda

I established nagar-iya

.nge' J and leas conven.

'<fa 'hitchhike-station';

- e.g. well-established

�paper)' - as well as

,boon 'teddy-bear'; and

'rming diminutives like

cloth', but also for a

a-it 'garlic-cheese', and

les for the soft-drinks

II these, a further note­

'n - One not addressed

, of prefixes based on

-le'umi 'inter-national'

an', tlat-memadi 'three-

," rlier stages of Hebrew.

" results. of our study

classic Semitic device

>n by means of an ac-

traditionalIy termed

'he theory of Natural

This was particularly

, focus on - Agents,

I is revealed by the

ed to innovate on the

Ifinitive form, that is,

itional stem-prefix as

well (compare, say, /e'exol 'to-eat'/axlan 'eater', lehartiv 'to-wet'/

marteva 'wetter'). Thus, responses on Part A, as summed up in (3)

below, overwhelmingly (88% in all) took some form of morpholog­

ical affixation - over one half in the form of root+incorporated

affixal pattern - i.e. Type (2) of the three listed above - and

around one-third in the form of stem/word plus external affix ­

which we had listed as potentially Type (I), or the most favored.

(3) Distribution of affixation devices in Part A coinages:

Noun Classes [Raw Boores]

Device �-1-1- Abet"-188 188

3S I 70 1599 [�53.5%J

lOS I 60 1395 [�35.3%J

-iya I .ut 1994 [�SS.S%l

AgtN_280

-....

Total %Ins"

'80

Root + Pattern 157 129205

Stem + Suffix 105 54 6S

[Most favorf"d

suffix] .an -iy&.an

Other devices, accounting for only 126 out of a total 1120 pos­

sible responses, included a small number of: blanks, real words

(conventional lexical items), conversion by means of benoni verbs,

compounds and blends. This tendency to rely on morphology, rather

than on conversion or compounding, was obviously also manifested

in Part B, where some form of affixation was the only option given

subjects. Yet even in this task, subjects clearly selected root + pat­

tern affixation far more commonly than they did stem plus external

affix options for the three classes of Agent, Instrument, and Place

included in this part of the test. Only with respect to Part C, the ..

listing of familiar words, were th;,re.ponsesvery mixed in this ­

respect. Yet here, too, some kind of formal affixal device was the

basis for many of the words subjects selected from their own voca­

bulary, applying to over half the Agent nouns, and around one­

third of Instrument (37%) and Place (33.5%) nouns. Only the Col­

lective and Abetract nouns tended to yield a more mixed, rather

inconclusive picture, as noted in Section 3.3 above.

These findings lead us to conclude that both in the well-established

or "old" wordstock, and even more 80 in their own innovative

coinages, Hebrew speakers make very broad use of three strictly

morphological, or word-internal, devices for word-formation: They

440

select root + internal affixes mainly in relation to familiar CaCaC

agent nouns and also for coining new CaCiC adjectives with the

'sense of '-able'; they rely heavily on combining consonantal roots

with affixal patterns which include a suffixal andjor a prefixal

syllable; and they increasingly tend to use word or stem plus an

external suffix for coining new names for a wide variety of semantic

classes of items - as noted at the beginning of this subsection.

3.5 Avoidance of Compoundinu

Perhaps the most striking result was that across the teat, subjects

avoided juxtaposition .. ar.compounding as a means of new-word

formation. The forms presented to subjects for selection in Part B

did not include any compounds; but in Part A only 5 out of more

than a thousand items were given in the form of a compound, while

in Part Conly 4% of the familiar words listed were compounds ­

mainly, as noted, for naming places, occasionally for instruments,

e.g. mxonat lcvi8ajdfuajtfira 'machine-for washing, printing, sewing'

respectively.' This accords exactly with the findings of our earlier,

oral study - where adults gave only 2.5% responses in compound

for innovative agents and instruments, even though half the input

verbs were presented to them together with a direct object (e.g.

"a girl whose job is to pull wagonB", "a tool that is used to scatter

buttonB". Clark & Berman 1984).

We choose to explain these findings as follows. Firstly, in strictly

8tructural terms, compounding in Hebrew is relatively limited, along

the following lines: It derives primarily compound noun8 from

nouns; it is restricted in the range ofcompound adjectivetl it allows ­

particularly in comparison with English and other Germanic lan­

guages (Meys 1975, Smith 1982); and it totally disallows compound­

v.ro formation, as is common in other languages (Clark & Clark

1979, Roeper & Siegel 1978, Mithun 1984 : 848) owing to the Semitic

constraint that all (although not only) verbs be constructed by

means of a fixed set of binyan conjugation patterns.

Secondly, in terms of actual usage, everyday spoken Hebrew, in

marked contrast to more normative formal styles of expository or

literary writing, deploys a variety of alternative 8trueture8 for

expressing noun-noun relations with no overt predicate. These in­

elude the widespread use of the gsnitive particle 8Ml 'of' to express

possession, and the tendency to substitute denominal-adjective

���

adjuncts for the l

in snch contexts I

(4) 8ixa telefon-it

talk phone-y

MneBrefu'i

congress medi

avoda mi8rad-i

work office-y

bgadi711lceyc-iy.

clothes summe

But over and a

syntactic and sem

tural options, we .

in current Hebrew

That is, speakers (

objects and entiti

community (DOWll

ent findings from t

revealing almost tc

for lexical innova1

temporary lexical' !

Hebrew speakers 1 \riving new words in I

catenating device ( i

a) In a task req' i,

30 compound expr \

one, the expreseion l

"like a single woro f

others received ar t'vacuum-cleaner = i

boker = 'morning-r \of the 30 expresaic

kova-plada 'steel h. \xevrat-bat 'daughtcll .

subjects to be very

speakers did not ,

"wordlike" in statu !

�II

!

relation to familiar CaCaC

:::aCiC adjectives with the

nbining consonantsl roots

suffixal and/or a prefixal

use word or stem plus an

a wide variety of semantic

ing of this suboection.

.t across the test, subjects

!lB a means of new-word

ts for selection in Part B

'art A only 5 out of more

arm of a compound, while

isted were compounds _

LBionally for instruments,

rELBhing, printing, sewing'

,e findings of our earlier,

� responses in compound

en though half the input

vith a direct object (e.g.

101 that is used to scatter

'llows. Firstly, in strictly

I relatively limited, along

compound nouns from

nd adjectives it allows _

,nd other Germanic lan­

.lly disallows compound­

nguages (Clark & Clark

48) owing to the Semitic

erbs be constructed by

patterns.

Tday spoken Hebrew, in

I styles of expository or

lternative structures for

ert predicate. These in­

.ticle shel 'of' to express

te denominal-adjective

>H

adjuncts for the more normative, classical form of noun plus noun

in such contexts as the following (and see n. 5): fi1,

!

(4) sixa tekfon-it VB

talk phone-y

MMS refu'i vs

congress medical

avoda misrad-it vs

work office-y

bgadim lceyc-iy-im vs

clothes summer-y

sixat telefon

talk- phone = 'phone conversation'

lcines rof'im

congress- doctors = 'medical meeting'

avodat misrad

work- office = 'office work'

bigdey Myic

clothes- summer = 'summer clothes'

But over and above these and other formal constraints - both

syntactic and semantic (Berman & Ravid 1986) - as well as struc­

tural options, we wish to suggest that compounding is not favored

in current Hebrew usage as a lexical device for new-word formation.

That is, speakers do not favor compounds as a means for labelling

objects and entities viewed as nameworthy within their speech

community (Downing 1977). As evidence, we note the very consist­

ent findings from the present study and from our earlier, oral study,

revealing almost total disregard of compounding as a possible option

for lexical innovation. And several other observations from con­

temporary lexical usage provide further support for our claim that

Hebrew speakers today prefer word-internal morphology for de­

riving new words in their language as opposed to the analytic, con­

catenating device of word-compounding.

a) In a task requiring native Hebrew-speaking subjects to rank

30 compound expression for relative degree of lexicalization, only

one, the expression yom-hulMet 'birth-day', was evaluated as being

"like a single word" by over 70% of the subjects, and only three

others received around 50% for this evaluation - sM'ev avak

'vacuum-cleaner = hoover', ke'ev 'fosh = 'head-ache', and aruxat­

boker = 'morning-meal, breakfast' (Berman & Ravid 1986). Many

of the 30 expressions listed there (e.g. xalom-balahot 'nightmare',

kooa-plada 'steel hat = helmet', pney hayam 'sea-face = sea-level',

xevrat-bat 'daughter-company = subsidiary') were judged by most

subjects to be very familiar, hence to some extent lexicalized. But

speakers did not construe such compound expressions as fully

"wordlike" in status.

1

r

b) Many of the lexicalized compounds which form part of the �

current Hebrew wordstock are the result of loan-translations taken �

over from languages rich in lexical compounds. These include nu- "merous everyday items such as those noted in (a) above, as

well as Mged-yam 'bathing-suit', mxonat-kvisa 'washing-machine',

xadar-sheyna 'sleeping-room = bedroom'. The external source of

such terms indicates that they are not the result of spontaneouB

coinages from within the monolingual Hebrew-speaking com­

munity, made by speakers who rely on their own native repertoire

of grammatical and lexical devices for new-word formation.

c) Alongside of such expressions as these, are many others which

were introduced as compounds by Hebraists early on in the revival

of the language as a spoken vernacular, but were subsequently

replaced by singleword items, derived by means of affixation, as

shown by comparing the earlier, compound forms in (5-i) with the

monolexemic forms currently in use in (5-ti).

(5) Agent (i) ish cava (ti) xayal

man-army soldier

Instrument te'udat masa dark-on

•certificate- travel way-Suff = passport

Place bet sfarim sifr-iya

house- books book-Suff = library

Dozens of examples could be added to these (as is done, for example

in Kutscher 1982, Sivan 1980). And there are many, many words

which might in principle have been introduced as compounds, where

morphological derivation was opted for - e.g. raftan 'dairyman'

f:om refet 'dairy', mavuna 'test-tube' from livxon 'to test', makOlet

'grocery-store', cf. kolel 'contains'. Interestingly enough, such

coinages rely on both types of affixation noted in Section 3.4 above:

Synthetic combination of affixal patterns with a consonantal root

and more analytic juxtapositioning of affixal endings to a word or

stem.

d) A fourth piece of evidence showing that speakers do not favor

compounding as a means for labelling objects and entities is pro­

vided by the phenomenon of clipping - e.g. the instrument nouns

mediax kelim 'washer-of dishes = dishwasher' and mazgan amr

'temperer-of air = air conditioner' are typically rendered by the

ioitial, head noun al

compounds, e.g. SUI

among many other!

styles, including po<

e) Another releva;

which both juxtap

morphophonologica

e.g. shmartaf 'wa

[fiverse] = 'limeric:

which ha;ve recentlj

(Berman & Ravid

peculiarly fused forI

given by over 100 •

gats where and ho'.

pounds (Berman, in

f) Next, we sugge

productive as a mel

are Ia;rgely restricte

a general-purpose,

various hyponomou

tionallexicon, Agel

compounding, in c(

M'al 'owner-of, ma;

language (and com

e.g. policeman, mai

Instrument and PI,

a quite restricted I

term such as mflX01l

'washing-machine' ,

'utensils' for colle<

'bedclothes', 'work

and, similarly bet­

head nouns in the ]

hospital'; 'sportsfie

respectively. These

compounds - as S'

novative agent and

olds ga;ve relativel:

pose, superordinau

cate agency (Clark!

----

which form part of the

of loan-translations taken

ounds. These include nu­

noted in (a) above, as

Tcvisa 'washing-machine',

The external source of

he result of spontaneous

Hebrew-speaking com­

air Own native repertoire

'-word formation.

" are many others which

ts early on in the revival

but were subsequently

means of affixation, as

d forms in (5-i) with the

ii).

:J;ayal

soldier

dark-on

way-Surf = passport

,ifr-iya

book-Suff = library

(as is done, for example

are many, many words

�d as compounds, where

e.g. raftan 'dairyman'

livzon 'to test', makOlet

astingly enough, such

ad in Section 3.4 above:

,ith a consonantal root

tl endings to a word or

t speakers do not favor

Jts and entities is pro­

. the instrument nouns

Iher' and mazgan avir

cally rendered by the

.�

initial, head noun alone. Even more striking are truncations of loan

compounds, e.g. super 'supermarket', teyp 'tape-recorder', tranziztor

among many others. And such clippings are also found in higher

styles, including poetry (Sadan 1979).

e) Another relevant phenomenon is the widespread use of a device

which both juxtaposes two words and fuses them into a single

morphophonological word in the form of blends in current Hebrew,

e.g. shmartaf 'watch-young = babysitter', :J;amshir 'five-verse

[fiverse] = 'limerick'. This process is very common with words

which have recently become entrenched in the conventional lexicon

(Berman & Ravid 1986, Nir 1980). Moreover, items taking this

peculiarly fused form account for some 15% of all the innovations

given by over 100 subjects in a test devised specifically to investi­

gate where and how Hebrew speakers do in fact form noun com­

pounds (Berman, in preas).

f) Next, we suggest that cases where compounding remains quite

productive as a means of constructing new lexical items in Hebrew

are largely restricted to a single type of semantic relation - where

a general-purpose superordinate term functions as the head, and

various hyponomous subordinates as adjuncts. True, in the conven­

tional lexicon, Agent nouns are typically not formed by means of

compounding, in contrast to the widespread use of the head-noun

M'al 'owner-of, master-of' for this purpose at earlier stages of the

language (and compare the many such compounds in English ­

e.g. policeman, mailman, milkman, doorman). On the other hand,

Instrument and Place nouns commonly take a compound form in

a quite restricted manner - with the head being a superordinate

term such as me:J;onat- 'machine' (cf. m:J;onat-kvisa, ktiva, gil'lia:l: for

'washing-machine', 'typewriter', and 'razor' respectively) or kley­

'utensils' for collectives (e.g. kley-mila, kley-avoda, kley-rh:ev for

'bedclothes', 'work-utensils = tools', and 'vehicles' respectively);

and, similarly bet- 'house-of', migrash- 'field-of', ulam 'hall-of' as

head nouns in the Hebrew equivalents of words meaning 'factory',

hospital'; 'sportsfield', 'tenniscourt'; or 'dance-hall', 'gymnasium',

respectively. These may be the most basic (or immature) kinds of

compounds - as suggested by the fact that in a task eliciting in­

novative agent and instrument nouns, English-speaking three-year

olds gave relatively many compound responses with general-pur­

pose, superordinate head nouns such as man, guy, woman to indi­

cate agency (Clark & Hecht 1982). And the relatively few compounds

ii

444

given on the same task by Hebrew-speaking children agelf 5 to 7,

(never by the three-year olds), mostly used the head noun mauhiT

'instrument' or mexona 'machine' (Clark & Berman 1984).

g) As a final BOurce of evidence for the fact that compounding is

not a common device for new-word formation in current Hebrew, l

we note evidence from research in progress on children's acquisition 1

and use of such constructions (Berma.n forthcoming. Bilev 1985).

A survey of children's usage in both interactive conversational

settings and in story-telling tasks reveals that relatively very few

compounds are used as part of the regular wordstock of these young

speakers. And we found virtually no innovative use of such terms.

at all, even in naming unfamiliar objects and animals in a story

picturebook, in contrast to the numerous within-word innovative

coinages occurring at this age (Berman & sag; 1981). Moreover,

by age 4 or 5, Israeli children do know how to form noun compounds

when required to do so in a structured elicitation task (Clark &

Berman in press). This suggests that preschoolers' natural or un­tutored knowledge of Hebrew includes the process of compounding

as part of the grammatical rules which they have internalized, but

that they do not necessarily deploy it as a spontaneous means offorming new words in their use of the language.

3.6 TTaMpaTency of Noun-Glaa8 Di8tinction8

The present study, as noted, extended an earlier investigation of

Agent and Instrument nouns, to include the categories of Place,

Collective, and Abstract nouns. We hypothesized that classes of

nouns which are semantically related might manifest a certain for­

mal similarity, too, whereas classes of nouns that cannot be sub­

.umed under a single superordinate category would take maximally

distinct surface forms. Thus, we expected names for Agents and

Instruments - as people and objects which perform activities _to share more surface forms than, say, Agents and Places. On the

other hand, we did not expect any strong pull towards total tTaM­paTency, or a fully one-to-one relation between meaning and form.

Such distinctiveness is often advocated by linguistic purists espe­

cially for purposes of self-conscious, official word-coinings. Forinstance, it has been recommended that the feminine patternmaCCeCa be used to label larger machines - e.g. makdexa 'pump,oHrig' - whereas masculine maCCeC be reserved for smaller, mainly

manual instruments

that the pattern C,

sabat 'porter, stev,

restricted to attril

811akran 'liar' (Rabir

of interpretation of

that familiarity wi,

embody violations ,

with knowledge of c

that conflict with D

meaning regularitie

Specifically, we

might share BOrne fc

(of Hebrew as of m

nahel 'manager' and

SOl" and potzan 'Cf

would share the maC

'sprinkler', maahtelt

would be formed pre

established yald-ut '

rather than with tl

vowel pattern with

'magic', Or the sff:

:twalon 'cessation'

COinages, such as �

Average percentag

selected (Part B)

of nOll

Device

CaCae

.an

Benoni

{verb participl

maCCeC

maCCeCa/et

miCCVC(afet)

." iya

'aking children aged 5 to 7

<sed the head noun ma:>:8hir

: & Berman 1984).

e fact that compounding is

nation in current Hebrew,

"" On children's acquisition

forthcoming, Bilev 1985).

interactive conversational

Is that relatively very few

. wordstock ofthese young

,ovative use of such terms

r,g and animals in a story

IS within-word innovative

& Sagi 1981). Moreover,

7 to form noun compounds

elicitation task (Clark &

�choolers' na.tural or UD-

, process of compounding

,ey have internaIized, but

a spontaneous means of

�uage.

"!8

n earlier investigation of

the categories of Place,

,thesized that classes of

t manifest a certain for­

illS that cannot be sub­

r would take maximally

names for Agents and

,h perform activities _

'nts and Places. On the

ull towards total trans­

sen meaning and form.

linguistic purists espe­

ial word-collings. For

the feminine pattern

- e.g. makdexa 'pump,

ved for smaller, mainly

.,.....

�

manual instruments - e.g. from the same root, makdeax 'drill'; and

that the pattern CaCaC be kept for occupational agents _ as insabal 'porter, stevedore', pasal. 'sculptor' - while CaCCan be

restricted to attributive terms - e.g. batlan 'idler, sluggard',shakran 'liar' (Rabinowitz 1947, and Ornal 1979 on the "regularity"

of interpretation of CaCaC nouns in Hebrew). We assumed, rather,

that familiarity with terms from the conventional lexicon whichembody violations of form/meaning distinctiveness would combine

with knowledge of common semantic generalizations to yield usagesthat conflict with more normative specifications for one-form/one­meaning regularities.

Specifically, we assumed that Agents and Instrument nOunsmight share BOme forms, as they often do in the established lexicon(of Hebrew as of many other languages) - e.g. present-tense me­nahol 'manager' and mehadek 'paper-clip', or CoCCan 10k/an 'aggres­sor' and pot:J:an 'can-opener'; that Instrument and Place nounswould share the maCCeCa pattern - as in well-established mamtera'sprinkler', mashtela '(plant) nursery'; and that Abstract nouns

would be formed primarily with an external out suffix _ as in well­established yald-ut 'child-hood', more recent manhig-ut 'leadership',rather than with the less unique, hence less transparent internal­vowel pattern with penultimate stress, e.g. tOhar 'purity', keaem'magic', Or the affixal pattern CiCaCon, e.g. shiga'on 'madness',xidalon 'cessation' - as is suggested by children's spontaneouscoinages, such as *cmi'ut for conventional cima'on lthirst-iness',

Table 5

Average percentage of innovative forms produced (Part A) andselected (Part B) and of worda listed (Part C) acr088 four classes

of nouns, using different s�ructu,ral devices

Dni"" I ,,-t I I�trom I P1a<e I 00Il�

CaCaO 8.5- - -.an

66.0 18.0- -Benoni

[verb participle] 17.0 12.5 --

maCCeC

-'-=-I� --

maCCeCa/et- 13.5 25.0

-miCCVC(a/et)- I 1l.O 33.5

-

... iya- - 24.0 I 58.0 I

446

*re'e1JUi for conventional ra'av 'hunger = hungriness' (Berman &

Sagi 1981).

This latter class proved problematic. Recall, firstly, that Abetract

nouns were not included in the judgement task in Part B of the

test; and in the production of coinages (Part A) as well as listing of

familiar words in Part B, responses were often in the form of

adjectives or else a mixed and varied, highly idiosyncratic set of

forms. The picture which emerges for preferred forms for the

remaining four categories, averaged across the three tasks of the

test - new-word formation in A, judgement of innovations in B,

and listing of occurrent words in C - is presented in Table 5 below.

The table liets only those devices which received as high as almost

10% of the total responses - so that the totals come to less than

100%.

The figures in Table 5 reveal a continuum of form/meaning in­

terrelations, as follows:

{61

i I I ( \

Agent Instrum Place Coaeetive

'----.-/ ��

At one end we find Age n t s, representing the most highly

"individuated" class of nouns, contrasting extremely with the less

specific class of Collectives at the other end. Ins t rum e n t

nouns are the most mixed subclass, sharing properties with both

animate Agents and statically located Places. This very clear find­

ing &CrOBB the three tests, such that Instrument nouns yielded the

most varied set of associatsd forms, accords well with flndings from

other studies (Clark & Berman 1984, Ravid 1978), as well as with

the set of instrument nouns in the conventional lexicon. Thus, of

all the classes examined here, they are the most highly restricted

in meaning: A person can be a carpenter, gardener, tennis-player

and dancer as well as a father, liar, braggart, or glutton at one and

the same time; but a scwors is nothing but a scwors, and as such

it is distinct from other objects also used for cutting, such as a knife,

a saw, a pruning fork, or a lawnmower. Lower down on the contin­

uum in (6) are Place nouns, which are semantically leaa restricted

in application than instruments, since a school, hospital, or factory,

say, can each be the location of numerous different activities, and

many different acts can be performed even in such specialized

places as a laundry, gymnasium, or restaurant. Besides, a place­

term is potentially ambiguous as between the location of an activity

or an object readin

for books, just as

plants. Hence at t

formal features w:

agents.

In an earlier stu

of a specific forma

domains (Berman

conventional choici

occupations in Het

(7) Device

CaCaC nagan

kanar

Benoni melavi

Verb meta!,

malxi'

-an p8ant.

Xacoc1

Loan muzik

Such facts from

findings of our Stl

meaning is only ps

collapsed into mol'

Agents, both OCCU]

for Instruments;. :

and -iya for both]

transparency - in

- applies to Mode

in principle pOBBes, .

word formation an '

tried to show abov

these different opti

In this final sect:

for Modem Hebre�

in general. Discuaai

��I

, hungriness' (Berman &or an object reading. Thus a library is both a place for reading andfor books, just as a nursery is both a place for planting and for

plants. Hence at the far end of the scale, Collective nouns shareformal features with place nouns, but not with instruments oragents.

In an earlier study, we pointed to the arbitrariness of selectionof a specific formal device even within highly restricted semantic

domains (Berman 1978: 394-401). This is clearly shown by theconventional choice of terms in such areas as music or educationaloccupations in Hebrew, thus:

,all, firstly, that Abstract

It task in Part B of the

.rt A) as well as listing of

'e often in the form of

�hly idiosyncratic set of

,referred forms for the

l the three tasks of the

�nt of innovations in B,

sented in Table 5 below.

,eived as high as almost

ootals come to less than

(7) Device Music Schooling

CaCaC nagan player ganen-et nursery-teacherleanar violinist

Benoni me/ave accompanist more teacherVerb metofef drummer mefalceax inspector

malxin composer marce lecturer-an psantran pianist targilan exerciser

xacocran trumpeter

Loan muzika'i musician st'ItdhU

profesor

1m of form/meaning in-

---,

Cotlective

�

.ting the most highly

'xtremely with the less

md. Instrument

: properties with both

I. This very clear find­

lent nouns yielded the

,ell with findings from

1978), as well as with

anal lexicon. Thus, of

nost highly restricted

trdener, tennis-player

or glutton at one and

, scissors, and as Such

:tting, such as a knife,

r down on the contin-

Itioally less restricted

, hospital, Or factory,

ferent activities, and

in such specialized

nt. Besides, a place­

-cation of an activity

Such facts from the conventional lexicon combined with thefindings of our study show that the identification of form withmeaning is only partial. and that such correspondences tend to becollapsed into more general superordinate categories, thus: -an forAgents, both occupational and attributive, and to a lesser degreefor Instruments;. mao ..a for both Instrument and Place nouns;and -iya for both Places and Collectives. Thus, avoidance of totaltransparency - in innovative usage ,,<, in the established lexicon- applies to Modern Hebrew, even though it is a language whichin principle possesses such a rich array of formal devices for new­word formation and one whose speakers in practice _ as we havetried to show above - still avail themselves liberally of many ofthese different options.

i!

i

1I

I4. DISCUSSION

In this final section, we consider the implications of Our findingsfor Modern Hebrew in relation to the iesue of lexical productivityin general. Discuesion of the role and nature of linguistic "produc-

448

tivity" is thus confined here to the lexical level of word-formation

processes aa against the creative or generative properties of rules

of the grammar. In the latter ClUla, operations and constraints are

structure-dependent, and refer to abstract structures such as N, V,

or A, Preposition or Particle, Subject or Direct Object, rather than

to words or c1aases ofwords . Lexical productivity, On the other hand,

concerns the extent to which a given word-formation device is

relied on'by speakers in construing the wordetock of their lan­

guages for purposes of interpreting both new and old words, aa well

aa for coining innovative terms. From this point of view, a lexica]]y

productive process is one which stilJ applies in the usage current

at a given point in the development of the language. Thus, use of

vowel-change to create causative verbs (as in pairs like rise-raise,

lie-lay, fall-fell) is nonproductive in English today, by contraat

with the commOn innovative use of a morphological device in the

form of the hif'il verb-pattern for this same purpose in current

Hebrew (Berman & Sagi 1981, Bolozky & Saad 1983); and the

Biblical pattern CaCiC (with a historica]]y long initial vowel) is no

longer productive for the class of agent-nouns in today's Hebrew,

by contrast with earlier words such as nagid 'governor', qacin

'officer' (glossed as explained in n. 4).

By these standards, however, this same surface pattern functions

as a "productive" process in Modern Hebrew, used for passive

'-able' adjectives - as was noted by Haim Blanc several decades

ago. Thus, the following are among the many such terms listed in a

conventional dictionary (Even-Shoshan 1979) as having been intro­

duced in modern times, which occur in actual usage: shavir 'break­

able = fragile',kari 'read-able = legible', kavis 'wash-able', shamish

'useable = practical', yasim 'applicable', amid 'stand-able = COn­

servable', pagia 'hurt-able = vulnerable', daliq 'burnable = flam­

mable', avir 'passable = traversable', qavil 'acceptable'. Nonethe­

less, this very "productive" means for attaching a given form to a

given meaning is not part of the grammar of Modern Hebrew, and

it remains a lexical phenomenon, much like its '-able' counterpart

in current English. Firstly, there are gaps in current usage, and not

a]] transitive verbs in fact form the baais for deriving passive ad­

jectives by this means - e.g. there is no word for 'approachable',

'disposable', or 'sendable' even though there are verbs from which

such adjectives could in principle be formed - unlike the theoreti­

cal form patir 'solv-able' which would be homophonous with a word

I,:

;:

i"1

with an unrelate,

slot that has aIr.

'one that is sen

adjectives in thi

zah,ir 'cautious' (

_ let alone all v

iu fact have this

Otber devices

meeting the reql

Hebrew include t

tiona" for coinin'

classes (Seetien :.,

fixes which we c

often in distinct

language (Section

Our findings ti

for innovations n

Cohen (1984) for,

Moreover, such cl tModern Hebrew, jloutset of this pap I,seemingly equall} I

and CaCCan. Th'l

Hebrew - e.g. (i,

malax 'seaman'. B \

many artizans, th [

such ouupations i

'gardener' , kacav 'I Iwas reserved in ] 1

attributes, mostly ( i

shakran 'liar', patj. }

however, some 10 • IMedieval Hebrew 1 \'hangman', kavra.1

And in contempor. i,

agent occupations, "

the CaCCan listed i \

been introduced ir tnames of attributel \.'crybaby') and of

,

i<,

I 2JIIP �

:callevel of word-formation

lerative properties of rules

rations and constraints are

,ct structures such as N, V,

Direct Object, rather than

uctivity, on the other hand,

word-formation device is

e wOrdstock of their lan­

new and old words, as well

is point of view, a lexically

plies in the usage current

;he language, Thus, use of

�aa in pairs like rise-raise,

.glish today, by contrast

Jrphological device in the

Bame pUrpose in current

& Saad 1983); and the

V long initial vowel) is no

lOUDS in today'B Hebrew,

nagid 'governor', qacin

surface pattern functions

ebrew, used for passive

n Blanc several decades

ny such tefD1B listed in a

79) as having been intro­

ual usage: shavir 'break.

avia 'wash-able', shamish

Imid 'stand-able = con­

daliq 'burnable = f1am­

.! 'acceptable'. Nonethe­

ebing a given form to a

)f Modern Hebrew, and

, itB '-able' counterpart

l current usage, and not

or deriving passive ad­

ord for 'approachable',

, are verbs from which

- unlike the theoreti­

10phonoUB with a word

with an unrelated meaning, or shaliax 'send-able' which would fill a

slot that has already been pre-empted by the agent noun meaning

'one that is sent, sendee = meBsenger'. Secondly, not all CaCiC

adjectives in this pattern (such as, say, Mishnaic zariz 'nimble'

ztiMI' 'cautious' or Modem samix 'viscous, thick', xadiah 'modem')

- let alone all words with this form in the established lexicon _

in fact have this meaning.

Other devices for word-formation which are characterizable aB

meeting the requirement of lexical productivity in contemporary

Hebrew include those which our test revealed to be "preferred op­

tions" for coining new nOUDS in a number of different semantic

c1asBoo (Section 3.1 above) and also the many stem-external af­

fixes which we characterized as highly favored in current usage,

often in distinct contrast to lexical norms at earlier stages of the

language (Section 3,4 above).

Our findings thus provide clea.r evidence that "speakers' choices

for innovations may shift over time" - as observed by Clark &

Cohen (1984) for agent suffixes in EngliBh, French, and Pomh.

Moreover, such changes may be particularly marked in the case of

Modem Hebrew, for sociohistorical reasons of the sort noted at the

outset of this paper. "Ve can illustrate this by reference to the two

seemingly equally productive agent patterns in Hebrew: CaCaC

and CaCCan. The former was used for agent nouns in Biblical

Hebrew - e.g. cayad 'hunter', dayag 'fisherman', tabax 'cook',

malax 'seaman'. By Mishnaic times, a period when the culture had

many artizans, this had become a common device for referring to

such occupations - e.g. nagar 'carpenter', zagag 'glazier' ganan

'gardener', kacav 'butcher'. The CaCCan pattern, on the other hand,

was reserved in Mishnaic Hebrew almost entirely for denoting

attributes, mostly ones with a negative import - e.g. batlan 'idler',

shakran 'liar', patpetan 'chatterbox' kamean 'miser'. Subsequently,

however, some 10 of the 35 or so CaCCan nouns introduced into theMedieval Hebrew lexicon are clearly occupation terms _ e.g. talyan

'hangman', kavran 'gravedigger', xacran 'yardman = janitor'.And in contemporary Hebrew, CaCCan forms are used in namingagent occupations almost as often as in naming attributes. Thus of

the CaCCan listed in the Even-Shoshan (1979) dictionary as havingbeen introduced in modern times, around the same number arenames of attributes (e.g. raxlan 'gossip', bazbezan 'wastrel', baxyan

'crybaby') and of occupations (e.g. raftan 'dairyman', rakdan

450

'dancer', canxan 'paratrooper') - the ratio being of about 6 to 5"

respectively. Moreover, in keeping with the increased reliance on

stem-external affixation noted in Section 3.4 above, Modern

Hebrew has also introduced several dozen agent terms formed out

of nouns plus the -an suffix - used occasionally for attributes

(e.g. tokf-an 'aggressor', harpatk-an 'adventurer', tlJU8t-an 'defeatist',

mahepx-an 'revolutionary'), but even more commonly to refer to

occupations (e.g. psantr-an 'pianist', xaliI-an, 'flutist', yecu'-an

'exporter', kaduragl-an 'football-er' and many others).

Thus, in contemporary usage, the CaCCan-pattern words and

-an ending words in general have come to represent the most

unmarked, least constrained means for naming agents - both

occupational like English farmer, sailor and attributive like liar,

idler. This specialized agentive functions of -an form words sets

them apart from CaCaC words in current usage in a way that dif­

fers from earlier, more classical norms. Although the CaCaC pat­

tern has yielded many new agent nouns - such as tayatJ 'pilot',

PalJal 'sculptor', kanar 'violinist', katav 'reporter' - these are COn­

fined to the occupation sense. And although speakers may often

come up with such well-established items when asked to cite agent­

nouns known to them (as was shown in the results on Part C, test­

ing retrieval from the conventional vocabulary, Section 3.3 above),

they will not themselves spcntaneously coin new nouns in this

form. One reason may be that this pattern violates the criterion of

distinctiveness (Section 3.6 above) by crOBBing lexical-class bounda­

ries, since CaCaC nouns are superficially homophonous with past­

tense verbs (ganav = 'thief' and 'stole', katav = 'reporter' and

'wrote') and they include many non-agent nouns (e.g. tavas 'peacock'

panas 'flashlight', pagaz 'mortar'). In much the same way, present­

tense or participial benoni forms, while also widely used for naming

agents in current as in classical Hebrew, are clearly not as semanti­

cally transparent or as specialized for the agentive meaning as are

nouns ending in -an. Thus, just as historically the CaCiC pattern

has shifted from classical agent functions to an adjectival '-able'

sense, so CaCCan and other -an ending forms are used today for a

wide range of agent meanings, both occupational and attributive,

and they are gaining an increased role in naming instruments as

well (Section 3.4 above). Hence, what constitutes a productive

word-formation device in encoding form/meaning relations may

differ considerably at different times in the history of a language.

.And this is not nece

bution of items in t

lauguage at a given

Against this bad

tion of lexical pre

introduced elsewhel

though interrelated

con": formal or str

spontaneous. Thus,

tura! devices and

grammar of a lang

and the formsl con

tions; 2) Nor m I

options favored for

by the Language 1

mains; the form-m

textbooks, and ot]

usage manuals; at:

construe as Hcarre(

derlies the spcntar

lexical gaps in the

nonspecialist users

tal settings, such a

study as well as ot ,

Ravid 1978).

With regard to

lowing. Firstly, it � I

most critically in al I

ing in Hebrew is I

3.5 above; the ma<

extendable to qua

underlies the nOU

xashben 'calculate'

fixal -n as a root c(

like *maxsheven to

as in targel 'to-exer

'practicer, exercise­

than the CaCCan

ale! or ayin low co

'announcer', but nc

ratio being of about 6 to 5

;h the increa.sed reliance on

ection 3.4 above, Modern

zen agent terms formed out

occa.sionally for attributes

enturer'. tVU8t-an ldefeatist'J �:

nore commonly to refer to

xa1il-an, <flutist', yecu'-an

many others).

;aCCan-pattern words and

me to represent the most

,r naming agents - both

. and attributive like liar,

's of -an form words sets

It usage in a way that dif­

Although the CaCaC pat-

s - such a.s tay"" 'pilot',

reporter' - these are con­

laugh speakers may often

: when a.sked to cite agent­

.he results On Part C, test­

mlary, Section 3.3 above),

. coin new nouns in this

" violates the criterion of

Ising lexical-cla.ss bounda­

homophonous with pa.st-

" 1catav = 'reporter' and

nouns (e.g. taV<L8 'peacock'

,h the same way, present-

.0 widely used for narning

re clearly not a.s semanti­

agentive meaning a.s are

ically the CaCiC pattern

I to an adjectival '-able'

oms are used today for a

)ational and attributive,

. naming instruments a.s

onstitutes a productive

'meaning relations may

.e history of a language.

.,.. c.,

Md this is not necessarily directly reflected in the numerical distri­

bution of items in the well-established, conventional lexicon of that

language at a given point in time.

Against this background, we can further refine our characteriza­

tion of lexical productivity by reference to the distinction we

introduced elsewhere (Clark & Berman 1984) between three distinct

though interrelated facets of the notion "productivity in the lexi­

can": formal or structural, normative or official, and colloquial or

spontaneous. Thus, 1) For m a I productivity refers to the struc­

tural devices and structure-dependent processes available in the

grammar of a language (in this case, as word-formation options)

and the formal constraints restricting the application of th-*le op­

tions; 2) Nor mat i v e productivity describes the structural

options favored for official purposes, such a.s: new words approved

by the Language Academy; terms coined in specific technical do­

mains; the form-meaning groupings listed in schoolgrammars, in

textbooks, and other pedagogic references; recommendations of

usage manuals; and the dev,ices which speakers self-consciously

construe a.s "correct"; while 3) Colloq u i a I productivity un­

derlies the spontaneous coinages evinced by speakers when filling

lexical gaps in the free flow of speech and the devices preferred by

nonspecialist users of the language in more structured, experimen­

tal settings, such a.s the kind described for Hebrew in the present

study a.s well as others (e.g. Bolozky 1978, Clark & Berman 1984,

Ravid 1978).

With regard to s t r u c t u r a I productivity, we noted the fol­

lowing. Firstly, it is here that the grammar and the lexicon interact

most critically in any language. Thus, for instance, noun compound­

ing in Hebrew is formally restricted along the lines suggested in

3.5 above; the maCCeC pattern for instrument nouns is not freely

extendable to quadriliteral roots - e.g. the verb xa8hav 'think'

underlies the noun maX8hev 'computer', but the related verb

xaBhben 'calculate' (rendered quadriliteral by addition of the suf­

fixal -n as a root consonant) is not the basis for an instrument noun

like *maX8Mven to mean 'calculator', just as the derived root t-r-g-l

a.s in targel 'to-exercise' does not yield *matregel or the like to mean

'practicer,exercise-device'; CaCaC agent nouns are more constrained

than the CaCCan pattern, aiDce they are avoided with root-final

ale! or ayin low consonantals - so that g-r- 2 'read' yields 1caryan

'announcer', but not *1cara (a fact that is shown by the currentgen-

452

eral reliance on the "incorrect" form cabay in place of normative!

caba from the root c-b-' for '[house]painter'); while stem plus -an

forma are less restricted than the root-incorporated pattern CaCCan

for denominal agent formation in the case of nouns with a more

complex structure than CVCVC - thus Xa.oet" 'yard' yields xacran

'janitor', retet 'dairy' yields rattan 'dairyman', but thefull noun­

stem is needed to derive nouns like mizrax-an 'orient-alist', tavru'a-n

'sanitation-ist'; and in structural terms of formal constraints, th�CaCiC '-able' pattern is not readily available for roots ending in a

glide - e.g. axil 'eat-able = edible' from the root 2 -k-l 'eat', but

not *shatiy 'drink-able' from the root shot-yo

Surprisingly enough, considering the rich tradition of morpholog­

ical research in Hebrew as in other Semitic languages, such issues

still await detailed investigation for Modern Hebrew, of a kind well

beyond the scope of the present study. One possible reason is that

concern with new-word formation in the language to date has fo­

cused mainly on what we have termed "normative" productivity.

Yet to the best of Our knowledge, little information is available on

the extent to which official recommendations for new-words have

filtered down into general use, to become part of the general

wordstock of Hebrew speakers. Exceptions are the studies of AI­

loui-Feinberg (1974) and Nir (1982) - both of which indicate that

in fact only part of the vocabulary that is officially instituted by

such a body as the Hebrew Language Academy is absorbed into

general everyday usage.

In the present context we willnote three phenomena in the lexicon

of colloquial Hebrew which run counter to what is sanctioned by

arbiters of "good" usage. Firstly, as observed in Section 3.5 above,

speakers often prefer Noun plus Denominal Adjective combinations

to the more classical construct-state forms of Head Noun plus Noun

Adjunct. This is true not only in a wide variety of syntactically

derived Noun-Adjective combinations, such as memshala yisra'el-it

'Israeli government' vs. the noun-noun counterpart memsMlet

yiB'ra'el (see Attiae 1981, Levi 1976), it is also manifested in the

recent tendency to create fully lexicalized compounds from such

strings - e.g. flvina Ivana 'white cheese', ta'asiya avirit 'aeronautical

industries', rake-vet taxtit 'nether train = subway'.

A second departure from normative dictates, as noted in Section

3.2 above, is the consistent preference of speakers for the maCCeCa

rather than miCCaCa pattern for Place nouns, acroBB a wide range

of different wor

Sh08han's (1979)

typically rendere

usage: mirpa'a 'e

d-r-k 'tread', mislshoP' from s-p-r

k-b-s 'launder', 11

mitpara 'sewing-'

s-p-n 'seaman', 1

'lawns' from dM;"

of. some _ three-dOl

The question of w

not immediately'

be fully "transpar

to the class of pIa<

instruments, say

mora 'pruuing-for

mamtera 'sprinkle] i

mavxena 'test-tub' :

and many others.

My analysis of .

argued above, spea "

meaning relations ( \note below, Hebre'

nouna, so can both

even though the I,

The second point .

simply may not kJ

overgeneralize to 1

Thus,alongsideoft

ered as maCCeCain

there are several wo:

form, which speak

'restaurant' and pe .

minkara 'tunnel', a

and there are othen

madflera 'hatchery',

'stairway'. Besides 1

the collectives mish

Occur alongside of

abay in place of normative t.1ter'); while stem plus -an

Jorporated pattern CaCCan i:aBe of nouns with a more f

Xacer 'yard' yields xacran iyman', but the fulI noun- "

Ii-an 'orient-alist', tavru'a-n

of formal constraints, the

lable for roots ending in a

n the root 2-k-l 'eat', buttoy.

ch tradition of morpholog­

itic languages, such iBBues

lrn Hebrew, of a kind welI

lne pOBBible reason is that

language to date has fo­

normative" productivity.

llormation is available on

tions for new-words have

)me part of the general

,ns are the studies of Al­

,th of which indicate that

is officialIy instituted by

cademy is absorbed into

phenomena in the lexicon

00 What is sanctioned by

ved in Section 3.5 above,

I Adjective combinations

of Head Noun plus Noun

variety of syntactically

,h as mem8hala yisra'el-it

counterpart mem8helet

I also manifested in the

i compounds from such

asiya amrit 'aeronautical

lway'.

�tes, as noted in Section

'eakers for the maCCeCa

lDB, across a wide range

of different words. Thus, the folIowing are alI listed in Even­

Shoshan's (1979) dictionary in the miCCaCa pattern, yet they are

typically rendered as maCCeCa words in unselfconscious, everyday

usage: mirpa'a 'clinic' from r-p-1 'treat', midraxa 'sidewalk' from

d-r-k 'tread', mishtala 'nursery' from sh-t-l 'plant', mispara 'barber­

shop' from s-p-r 'cut (hair)', mixbasa 'laundry, washroom' from

k-b-s 'launder', mishxata 'slaughterhouse' from sh-h-t 'slaughter',

mitpara 'sewing-room' from t-p-r 'sew', mispana 'shipyard'from

8-p�n (seaman', mizraka 'fountain' from z-r-q 'throw', midsha'a

'lawns' from deshe 'gr8BB' (and these represent only the commonest

of some three-dozen such neologisms listed in this dictiol'ary I).

The question of why the normative form is resisted in such cases is

not immediately obvious, particularly as this pattern would then

be fully "transparent", as follows: It would be uniquely allocated

to the class of place-nouns, and maCCeCa would serve primarily for

instruments, say - as in older, Biblical maxresha 'plough', maz­

me:ra 'pruning-fork', Medieval magreja 'rake' J and also recent

mamtera 'sprinkler' as well as (non-agricultural) implemente like

mavxena 'test-tube" maclema 'camera', masrega 'knitting-needle',

and many others.

My analysis of the situation is as follows: Firstly, as we have

argued ahove, speakers are quite tolerant ofnon-uniqueneBB in form­

meaning relations of this type. Just as English -er (and, as we shall

note below, Hebrew meCaOeG) serve for both agent and instrument

nouns, so can both miCCaCa and maCCeCa serve for place nouns,

even though the latter is also commonly used for instruments.

The second point relates to a further kind of opacity. Speakers

simply may not know when to use which form, and hence they

overgeneralize to the leBB speciaIized, less restrictive maCCeCa.

Thus, alongside of the words listed as iniCCaCa but generally rend­

ered as maCCeCa in ordinary speech - such as those noted above ­

there are several words which have become fOBBilized in the miCCaCa

form, which speakers never change to maCCeCa (e.g. mis'ada

'restaurant' and perhaps by direct analogy mizlala 'glutton-ery',

minhara 'tunnel', and also mixlala 'college', midrasha 'seminar');

and there are others which are rendered only by maCCeCa - e.g.

madgera 'hatchery', mazleva (normative maxlava) 'dairy', madrega

'stairway'. Besides there are nonplace nouns in both forms - e.g.

the collectives misktara 'police', makhela 'choir'. And older words

OCcur alongside of more recent coinages in all subgroups I This

suggests that there is no motivated, morpho-phonological or se,

mantic basis for speakers to make a decision as to which form is

"right" in the sense of best suited to their own norms of usage.

Some words have become lexicalized one way, others another, on

the basis of common usage rather than of normative dictates or

structural constraints. Where no such fossilization has occurred,

speakers will either opt for normative miCCaCa or they will extend"

the less specialized maCCeCa to place-names in accordance with

their individual lexicon, as a function of their personal linguistic.1

history and experience. Thus, it is precisely in such instances, I

where "colloquial" and "normative" productivity tend to conflict, !that lexical divergence-and variation can be expected. I

The last set of instances we note here is of the masculine-noun I

pattern maCCeC. Since Mishnaic time, this has come to be more

specialized for the instrument sense - as in masrek 'comb', mash.pex I

�:::;; �:::;.�m:7b�B':��:�;�����:k���:)��:;o:�;�; Ithis pattern is typically specified as the form par excellence for I

naming instruments in Hebrew. Yet our studies indicate some re- ,sistance to this normative recommendation. Thus, this pattern !

was rarely used in subjects' production of innovative instrument I

nouns in the oral test (Clark & Berman 19B4); it constituted only

21% of the answers in the comparable written test (Part A of the

present study); and less than a third (32%) of the forms chosen as

suited to instrument nouns (Part B) were in the maCCeC form. We

suggest that the official requirement is successful in the case of

words that can be defined as rote-learned or as unanalyzed at two

extremes of the Hebrew wordstock: Ones which are common, every­

day terms that form part of the basic vocabulary of young Hebrew

learners, and so have become fossilized in this set form (as illustrated

in (B-i) below) and those which are part of the highly specialized,

selfconsciously innovated technical terminology of the language

(as in B-iii). Elsewhere, as in the examples in the middle column

(B-li) below, speakers quite typically use a present-tense benoni

verb-form - most particularly where the base-verb is in the P3

pi'el pattern, rather than in one of the two other transitive patterns

PI pa'al and P5 hit'il.

(8) Instrument Nour

(i)

Rote-Learned, FOEl

mazleg < PI",

fork d

mashpex < PI s}

funnel p'

malt'ax <PIp.

key cJ

masrek <P3m

comb c<

mavreg <P5 m

screwdriver S(

of. b6reg 'screw'

m'

ba,

mt.,

re<

m,

trE

m,

an

m<

cal

The first set of we

Jished items known

Hebrew as in other I

"" isolated items, wi

verb. The last colum

people particularly f

norpho-phonological or 00. fecimon as to which form is

I their own norms of usage. •:l6 way, others another, on i'

; 1 of normative dictates or :

f088ilization has occurred,

, iCCaCa or they will extend .

names in accordance with

)f their personal linguistic

ecisely in such instances, ,

oductivity tend to conflict,

l be expected.

I is of the masculine-noun

�his has come to be more

in maarelc 'comb' J maahpex

like maBfler 'metalworker',

Huska 1981). And today

;, form par excellence for .

studies indicate some re­

,tion. Thus, this pattern

of innovative instrument

[984); it constituted only

ritten test (Part A of the

{,) of the forms chosen as

in the maCCeC form. We

succe88ful in the case of

or as unanalyzed at two

: ,hioh are common, every­

: ,bulary of young Hebrew

,lisootform (as illustrated

, ,f the higWy specialized,

nology of the language

,s in the middle column

a present-tense benoni

, base-verb is in the P3

)ther transitive patterns

.,...

(8) Instrument Nouns in the maCCeC Pattern:

(il

Rote-Learned, Fossilized

�

(ii)

Innovative, Resisted

mazlefl

fork

mashpex

funnel

mafteax

key

masrek

comb

< PI zolCfl

drips

< PIshofex

pours

< PI poleax

opens

< P3 mesarek

combs

makrer

fridge

maghec

iron

mafceax

nutcracker

maxshev

computer

maxded

sharpener

�P3 mekarer

cools

� P3 mCflah(J(;

presses

�P3 mefadax

crackc

� P3 mexaBhev

calculate

�. P3 mexaded

sharpens

The first set of words are among the high-frequency, well-estab­

lished items known to be resistant to change or regularization inHebrew as in other languages (Schwarzwald 1982); they are learntas isolated items, without any analysis relating them to the base­

verb. The last column consists of technical terms, typically used by

people particularly familiar with the referents in question - and

i

mavrCfl < P5 mavriu

screwdriver screws

cf. b6rCfl 'screw'

(iii)

Specialized, Frozen

macber < PI cover

battery cumulates

maklet < PI kola

receiver absorbs

maBhder < P3 meshader

transmitter transmits

magber <P5 magbir

amplifier increases

macmed <P5macmid

car-clutch links

456

they indeed represent the result of selfconscious, official, policy­

making ofthe kind we have described as "normative productivity".

The middle set is perhaps the most interesting - since the forms to

the left are those officially recommended, either originally or to

this day (the words for 'refrigerator' and '(pencil-)sharpener' have

been standardized in the "deviant" form) and there are others

which could be added to this list, e.g. masnen 'filter', commonly

rendered as mesanen.et '(kitchen) sieve'. These words are instances

where speakers have opted for the less transparent, non-unique i

device of conversion - retaining the present-tense participial verb­

form which is very close in pronunciation to the maCCeC form, so

that the same surface form is used both as a present-tense verb and

as an instrument noun, as shown in (8-ii) above.

Thus, although the ma- prefix nouns are "taught" in school­

grammars as the class of words for naming instruments, although

a standard dictionary lists well over 100 such nouns as having en­

tered the language recently, and although - as noted in Section

3.2 - speakers are selfconsciously aware that this is the "good" or

"correct" way to derive instrument nouns, actual usage may run

counter to these dictates. Even people who work with computers

often name them by the present-tense plural form mexashv-im and

not by "required" maxshev-im, just as people who work with cars

are likely to call the radiator either by the loan-form radiyator or

by the present-tense form mecanen 'chills/chiller' in preference to

normative macnen (and soo, further, Alloni-Feinberg [1974] for a

sociolinguistic study of the gap between official nomenclature for

car-parts and actual usage in different sectors of the population).

In fact, if someone talks about makrer when referring to a refrige­

rator, say, in preference to colloquial present-tense mekarer, he or

sue is likely to be identified as a schoolteacher, a grammarian, or a

foreigner. There is thus ample evidence that speakers are resisting

the maCCeC form as "bookish", except in the more selfconsciously

monitored contexts of technical expertise.

Where colloquial usage conflicts with normative dictates - as

in use of maCCeCa for miCCaCa place-terms, and in use of meCaCeC

for maCCeC instruments - two related trends emerge. There is a

pattern of considerable variability across speakers and even within

a single individual depending on the context of usage, Whether for­

mal, hence more selfconscious, or casual and hence le88 monitored.

And language change can be predicted, as a given, non-normative

set of forms becon

a new "standard"

The question reI

devices comes to

concerns us here.

stnidural options a

- which is how w

pounding and the

Another structural

we took to explain

name agent nouns.

find expression in

enee, even in a Ian

in Hebrew. A furth,

that speakers will I

common in their 1

itself requires claril

listed in a conventi '

language as well '" '

classes of Agent, In

have shown this to

ers themselves. Th

certain form many

speakers at a parti,

In that case, the d,

as a basis for prod'

issue here.

We conclude tha

word.formation dev

relevant devices fav

These will depend ,

(i) Underlying stru '

available to speaker;

pattern incorporati( �

as suffixing in Eng

affixation in Semit

(iii) distribution of

Jished lexicon, with

their everyday disc

frequencies establisl

,onscious, official, policy­

'normative productivity".

•ting - since the fornls to

d, either originally or to

'(pencil- )sharpener' have

m) and there are others

�asnen 'filter', commonly

rhese words are instances

transparent, non-unique

3nt-tense participial verb-

I to the maCCeC form, so

a present-tense verb and

above.

are "taught" in school­

ng instruments, although

such nouns as having en­

h - as noted in Section

that this is the "good" or

lS, actual usage may run

'ho work with computers

Iral form mexashv-im and

,ople who work with cars

he loan-form radiyalar or

'/chiller' in preference to

mi-Feinberg [1974J for a

official nomenclature for

lctors of the population).

nen referring to a refrige­

,sent-tense me1carer, he or

wher, a gramma.rian, or a

hat speakers are resisting

1 the more selfconsciously

normative dictates - 8S

lB, and in use of meCaCeC

orends emerge. There is a

speakers and even within

3xt of usage, whether for­

md hence less monitored.

B a given, non-normative

.,.. ,

set of forms becomes established in general usage, hence reflecting

a new "standard" (Donag-Kinrot 1978) .

The question remains as to why or how a given device or set of

devices comes to be productive in the "colloquial" sense which

concerns us here. One factor may be general favoring of certain

structural options at a given phase in the development of a language

_ which is bow we explained, for instance, the avoidance of com­

pounding and the wide use of stem-external affixes noted earlier.

Another structural factor may be the pull to di8tinctivene88 - which

we took to explain the current preference for -an ending words to

name agent nouns. However, as we have pointed. out, this will never

find expression in anything like a total form-meaning correspond­

ence, even in a language which affords the varied options available

in Hebrew. A further factor is that of frequency, since it is reasonable

that speakers will make broadest use of those forms which are most

common in their language. However, the notion of "frequency"

itself requires clarification. If it refers to the established wordstock

listed in a conventional dictionary, and covering all periods of the

language as well as all levels of usage, then our findings for the

classes of Agent, Instrument, Place, Collective, and Abstract nouns

have shown this to be incompatible with the choices made by speak­

ers themselves. There may be dozens or hundreds of words in a

certain form many of which are not known, or not used at all, by

speakers at a particular point in the development of the language.

In that case, the devices which they embody cannot be considered

as a basis for productivity in new-word formation of the kind at

issue here.

We conclude that in characterizing the relative productivity of

word-formation devices, account needs to be taken of the currently

relevant devices favored by members of a given speech community.

These will depend on a complex interaction of factors, including:

(i) Underlying structural welIformedness and the formal options

available to speakers - e.g. vowel alternation and root plus affixal

pattern incorporation in Hebrew; conversion and prefixing as well

as suffixing in English; (ii) typological predispositions - e.g. for

affixation in Semitic languages, for compounding in Germanic;

(iii) distribution of these devices in the conventional, well-estab­

lished lexicon, with the vocabulary items employed by speakers in

their everyday discourse often differing quite considerably from

frequencies established for written texts; (iv) psycholinguistic fac-

458

tors favoring distinctiveness and semantic as well as morphological

transparency; (v) speaker expectations deriving from the individ­

ual's experience, level of literacy, background in formal language

study, and personal norms of usage; and (vi) patterns of change,

of regularization, and extension of various devices at a given point

in the historical development of a language. From these several

points of view. Modern Hebrew seems to afford a particularly in­

teresting ca.se for investigation. It offers a rich array of affixal word­

formation devices ranging from the highly synthetic to aggluti­

nating type; owing to its having been so recently, and quite uniquely.

revived a.s a spoken vernacular and the official language of a partic­

ular gee-political entity,- words are constantly being innovated to

name entities unfamiliar from prior stages of the language; and

there is a peculiar tension in the society between the structurally

motivated pull to regularization of form/meaning correspon­

dences and the conservative reliance on earlier, written sources

on the part of official innovators and normative grammarians, on

the one hand, and the rather different motivations and sources

of hypothesis construction relied upon by speakers both when

acquiring and when using the language as an everyday means

of expression.

Address of the author: Ruth A. Berman

Department of Linguistics

Tel Aviv University

Ramat Aviv

Israel 69978

NOTES

1 The study l"eported on here forms part of a broader pl'ojeet in the general

domain of word.formation, including croBSlinguistio research into ehlldren's

development of word�formation devices. I am indebt.ed to Professor Eve

V. Clark of Stanford University and to Dorit Ravid of Tel.Aviv University

for their assist,anee, and for providing invaluable insights on all phases of

this work.

11: In representiI� Hebrew forms, both current and classical, a broad

phonetic transcriptIon is adopted, as a rough rendering ofhow such items are

pronounced in what Blanc (1954) termed "General Israeli Hebrew". Thus, we

do not replicate the historical (or orthographic) consonantal root elements,

unless these are relevant to a particular line of argument. Words have final

stress, unless marked by an accent aigu as having penultimate word-stress.

In representing the morphological affixation patterns termed tniRhkal

'weight' for nouns and adjectives, binyan 'construction, conjugat.ion' for

verbS, we adopt thf .

consonant, other ele i

sarak 'combed', samstands for tisroket 'ha

while maCCeC st-and; Iand miCCaC for mia �

3 The study was c ;

department at Tel-A;

indebted to Miriam : lMendelewitch and Be !

4 Words are glossl; i

�e, even though i !earher stages of the 1 \Miahnaic, rabbinical l'Hebrew compow l

that order, with the !C'construc:t-stat..e") f jconstituents in Nom

followed by modifier'number and gender ',�

e�tion. we transll !directl:y into EngliBh I

6 ThIS is manife8te I

same CaCCan-patterr I

together' for both the [.

the recent Instrumen fto improving Hebrew i1982), list.eners were j Ithe maCCeC pattem,1

Yet when I asked spel:: I

was very often to ase 'Ihistorical pharyngeal I.something akin t.o 'm ­

tive dictates are not Ialthough the two seql '

to this day, in most l 1

AlIoni-Feinberg, Yafa IA stuny of knowlec rAdvance8 in Langu. t

Aronoff, Mark. 1976. \M. 1. T. Press. ,

AttillB, Talia. 1980. TJ ITel-Aviv Universit. 'I

Attias, Talia. 1981. (University master': f

Avinery, Yitzchak. 19 [b.ra'el Publishing. \

Balgur, Raphael & Me I

Tel-Aviv: Sifrivat ] IBarkali, Shaul. 1964. ( ,Berman, Ruth. 197t,

Publishing Project! IBermlUl, Ruth. 1982.: •of productivity. Pal'English, Bar-Han 1 f

c as well as morphological

deriving from the individ_

round in formal language

I (vi) patterns of change,

s devices at a given point

lage. From these several

, afford a particularly in­

rich array of affixal word­

>ly synthetic to aggluti­

ently, and quite uniquely,

'iciallanguage of a partie­

.antly being innovated to

;es of the language; and

between the structurally

orm/meaning correspon-

earlier, written sources

rmative grammarians, on

motivations and sources

by speakers both when

as an everyday means

gllstics

Y

)ader project in the general

tic research into children'sndebt.ed to Professor Eve

vid of Tel-Aviv University01 insights on all phases of

:it and classical, a broad

3ring of how such items are

Israeli Hebrew". Thus, weonsonanta.l root elements,

'gument. Words have final

� penultimate word-stress.

patterns termed miflhkal

truction, conjugat,ion' for

.,.. ,

verbs, we adopt the convention of indicating radical elements bv C for

consonant, other elements by phonetic segments, e.g. CaCaO stands forsarak 'combed', saman 'marker', and kat-av 'wrot-e', 'reporter'; tiCC6Cet

stands for ti8r6ket 'hairdo', tism6net 'symptom', and tixt6vet 'correspondence'while maCCeC stands for masrek 'comb', meCuCaC for me8Uman 'marked',

aod miCCaC for mi.xtav 'letter'.J The study was conducted as part of a class projPct in the linguil:'tics

de�ment at Tel-Aviv University during the 1982/83 school year. I am

indebted to Miriam Saar for help at all phases of the study, and to Anat

Mendelewitch and Sonia RaCf for providing part of the data.

"Words are glossed according to their most accepted sense in current

usage, even though in many cases they harl a rather different meaning at

earlier stages of the language, whether in classictil Biblical Hebrew or later

Mishnaic, rabbinical Hebrew and Medieval writings.

5 Hebrew compounds take the form of Head Noun + Adjunct Noun in

that order, with the initial, head noun often in a morphologically bound

("construct-state") form, distinct from its free, nongenitive form. The

constituents in Noun + Adjective phrases occur in the same order, head

foHowed by modifier, but then the adjective agress with the head noun in

number and gender - as in the examples of (4) of the text. For ease of

exposition, we translate both noun-noun and noun-adjectives combinations

directly into English.

• This is manifested for instance in the colloquial tendency to use the

same CaOOan-pattern word sha.dxan from the verb meaning 'connect, tie

together' for both the older, well-established, Agent noun 'matchmaker' und

the recent Instrument term 'stapler'. On a duily radio program dedicated

to improving Hebrew usage (riga shel ivrit 'fA] Moment of Hebrew', May 19,

1982), listeners were instructed to use 8 distinct term for the instrument, in

the maCCeC pattern, thus: maalev from the noun ldiv, the pin used in stapling.

Yet when I asked speakers how they would interpret this word, their responsewas verv often to associate it with the word xalav 'milk' - with an initial

historic8.l pharyngeal Uchet" - so that they interpreted the word as meaningsomething akin to 'milker', 'milk-machine'. This clearly shows that norma.

tive dictates are not always coneistent with formal transparency - for

although the two sequences of x.l.v are distinct in the Hebrew orthogrtlphy

to this day, :in most current pronullciation they sound identical.

REFERENCES

ABoni-Feinberg, Yafa. 1974. "'Official Hflbrew terms for parts of the cal':A study of knowledge, usage, and attitudes." In: JOShua A. Fishman, ed.Advance8 in Language Planning. The Hague: Mouton, 493-517.

Aronoff, Murk. 1976. Word Formation in Generat��ve Grammar. Cambridge:M. 1. T. Press.

Attias, Talia. 1980. The fonnation of diminutives in contemporary Hebrew.Tel-Aviv University IDS.

Attias, Talia. 1981. Ordering of Adjectives in Modern Hebrew. Tel-AvivUniversity master's thesis.

Avinery, Yitzchak. 1976. A Thesaurua 0/ the Hebrew Radical Nouns. Tel-Aviv:17.ra'el Publishing.

Balgur, Raphael & Menachem Dagut. 1975. Ba.sic Hebrew-English Dict'ionary.Tel-Aviv: Sifrivat Maariv. [Hebrew].

Barkali, Shaul. 1964. Oomplete Noun Table. Jerusalem: Rubin :Mass [Hebrew].Berman, Ruth. 1978. Modern Hebrew Structure. Tel-Aviv: Universit,y

Publishing Projects.

Berman, Ruth. 1982. New-word formation in English and Hebrew: The issue

of productivity. Paper. p�ntcd at Conference of University Teachers ofEnglish, Bar-Ilan Umver8lty, Ramat-Gan, June 1982.

Berman, Ruth in press. "The role of blends in Hebrew word-formation".

Med.uerranean Language Review.

Berman, Ruth. Forthcoming. A developmental route: Learning the form and

function of complex nominal8 in Hebrew.

Berman, Ruth, Barbara F. Hecht & Eve V. Clark. 1982. "The acquisition of

agent and instrument nouns in Hebrew.'� Papers and Reporta on Ohild

Lan{fUa{/e Development 21, 16-24.

Berman, Ruth & Dont Ravid. 1986. "Degrees of lexicalization in Hebrew

compounding." Hebrew Oomputational Linguiatic8 Bullet1.n 24,5-22

[Hebrew]' !

B�rmanJ �uth & Yisrael �agi. l��l. "Word-format�on pt'O?eBS�S �d lexic�l fmnovatlOns of young children. Hebrew Computational LtnguUJtu:.s Bullet�n f­

18, 36-62. rHebrew]. 'Rilev, Roni, 1985. The development and use of noun Compounds by Hebrew- 0

speaking children. Tel-Aviv University School of Communications Dieor- �

ders M. A. thesis. [Hebrew]. �

Blane, Haim. 1954. "The growth of Israeli Hebrew." Middle ErMtern Affairs i

5, :185 -392. !Bolozky, Shmuel. 1978. uWord-formation strategies in the Hebrew verb �

system." Ajro.Asiatic Linguistics 5, 1-26. fBolozky, Shmuel & George Saad. 1983. "On active and non-active cRusativiz- f

able verb in Arabic and Hebrew". Journal of Arabic Linguistics ]0, 71-80. t

Clark, Eve V. & Ruth A. Berman. 1984. "Structure and use in the acquisition Iof word fonnat,ion." Langua.ge 60, 542-590. .

Clark, Eve V. and Ruth A. Berman. In press. "Types of linguistic knowledge:

Interpreting and producing compound nouns." Journal of Ohild Language. i

Clark, J.�ve V. & Herbert H. Olark. 1979. "When nouns surface as verbs." .

Lanrruage 55, 767-811.

Olark, Eve V. & Sophia R. Oohen. 1984. "Productivity and memory for

newly.fomled words".•Journal oj Ohild Language 11, 611-626.

Olark, Eve V. & Barbara F. Hecht. 1982. "Learning to coin agent and

instrument nouns". Cognition 12, 1-24.

Donag-Kimot, Rina. 1978. The Language of Students in Israeli Sohools:

Language Usage. Standard. and Norm in the Language of Native.Born

ISl"ueli Students. Jerusalem: Hebrew University doctoral dissertation

[Hebrew].

Downing, Pamela. 1977. "On the creation and use of English compound

nounR." Language 53, 810-842.

Dress]m', Wol(lSang U. 1981. "On word formation in Natural Morphology."

Wi,ner Lingui8!i,sche Gazette 26, 3-13 (= PICL 13, 1983, 172� 182).

Du Nour, Miriam. 1979. "Form and meaning in nouns and adjectives eurling

in .an, -ani." In: Rabin and Fischlflr, eds, 37 -41. [Hebrew].

Even-Shoshnn, Avraham. 1979. The Concentrated Hebrew Dictionary. Jenl.

salem: Kiryat Sepher. [Hebrew].

Gluska, Yitzchak. 1981. "Nouns ofthe maqtel pa,ttern in Biblical and Mishnaic

Hebrew." Leshonenu 45, 3/4, 280-298. [Hebrew].Kutscher, Eduard Yeche7'.kel. 1982. A History of the Hebrew Language, ed. by

Raphael Kutscher. Jerusalem: Magnes Press; Leiden: Brill.

Levi, Judith N. 1976. "A semantic analysis of Hebrew <'-ompound nominals."

In: P. Cole, ed. Studies in Modern Hebrew Syma.'"C and Semantic.'f. Amsterw

dum: North-Holland, 9-55.

Meyl'l, W. J. 1975. Compound Adiecti'008 in English and the Ideal Speaker/Lis.tene.r. Amsterdam: North.RoHand

MithuH, Marianne. 1984. "The evolution of noun incorporation." Language

60, 847 -894.

Nahil', Moshfl. 1978. "Normativism and educated speech in Morlern Hebrew."

Internalionvl J01trnal 0/ the Socwlogy of Language 18, 49-67.

Nil', Raphael. 1980. "The semantic structure of nominal compounds in

Modern Hp,hrew." Bulletin 0/ the Sclwol oj Oriental & African Studies43, 185-196.

Nit, Raphael. 1982, U) .Academy of the HEBuU<tm 19, 20-33.

Oman, Uzzi, 1979. "0Hebrew verb." In I

Rabin, Chaim. 1984. II,R. Cooper, ed. Sooedition of: lnternati<

Rabin, Chaim &Ben.ZiStudies in Hebrew at

Teaching of Hebrev

Rabinowitz, Solomon.

ed. Daniel Persky. :Ravid, Dorit. 1978. We

Adjectives. Tel.AvhRavid. Dom. In prelA Study of Morpho!versity doctoral dis:

Roeper, Thomas & McomJX!unds." Lingtt'

Sadan, Dov. 1976. U(Ben-Zion FishIer &Jerusalem: Council

SchW8rZWald, Ora R. ]Gan: Bar-nan Unb

SchwarzwaId, Ora. 191BeOhinuch 35, 163.

Sivan, Reuben. 1980Rubinstein.

Smith, Karl. 1982. Th'Norwegian. Tel�Avi

Werner, Fritz. 1982. '"16, 263-295.

Werner. Fritz. 1983.1Harrassowitz.

------yr '! i

'" ���;'.I'C\\ WUnl-l0!'lllat1on".

u route: Learning the form and

;lark. 1982. "The acquisition of' Papers and Reports on Ohild

�e8 of lexicalization in HebrewLinyuiBtics Bulletm 24,5-22

ormation processes and lexicalnputational Linguistics BUlletin

:- noun Oompounds by Hebrew_ I001 of Communications Disor_

brew." Middle ,Ea.stern Affairs

.rategies in the Hebrew verb

jive and non-active causstiviz.Arabic Linguistics ]0, 71-80.ture and use in the acquisition

rv>es of linguistic knowledge:,. 'JournalolOhild Langwige.aen nouns surface as verbs."

roductivity and memory forruage ll, 611-626.

Learning to coin agent and

i. Students in Israeli Schools:Ie Langnage of Native-Born'emity doctoral dissertation

d use of English compound

;1 on in Natural Morphology."PIOL 13, 1983, 172-182).nouns and adjectives enrling-41. [Hebrew].

�d IIebrew Dictionary. Jern-

!;ern in Biblical and Mishnaic'raw].the Hebrew Language. ed. byLeiden: Brill.

,brew compound nominata."�1:' and Semanticlf. Amster_

, and the Ideal Speaker/Lis_

1 incorporation." Language

Jpeech in Modern Hebrew."age 18, 49-67.

)f nominal compounds in)rientaJ. & African Studies

:Nit Raphael. 1982. l'Linguistic transparency of neologisms coined by theAcademy of the Hebrew Language." Hebrew Oomputational LinguisticaBullet'" 19, 20-33. [Hebrew].

oman, Uzzi, 1979. "Once more: The meanings of the conjugations of theHebrew verb." In Rabin & Fischler, OOs, 11-19. [Hebrew].

Rabin, Chaim. 1984. "The 8OC1iology ofnormativism in Israeli Hebrew." In:R. Cooper, ed. Sociolinguiatic Per8PectiV68 on Israeli Hebrew (Specialedition of: International Journal 01 the Sociology 0/ Language).

Rabin, Cheim &Ben.Zion Fischler. (ads) 1979. SlUomo Kode8hJubilee Volu,,",:Studie8 in Hebrew and the Teaching of Hebrew. Jerusalem: Council on theTeaching of Hebrew. [Hebrew].

Rabinowit:t.:, Solomon. 1947. SeIer Hamiskalim: Studies in Hebrew Philology,ed. Daniel Persky. New York: Shulsinger. [Hebrew].

Ravid. Dorit. 1978. Word-Formation Processes in Modern Hebrew Nouns andAdjectives. Tel-Aviv University master's thesis.

Ravid, Dorit. In preparation. Transient Phenomena in Child Language:A Study of Morphological Usage in Different Age.Groups. Tel�Aviv Uni­versity doctoral dissertation.

Roeper, Thomas & M. Siegel. 1978. "A lexical transformation for verbalcompounds." Linguistic Inquiry 9, 199-260.

Saden, Dov. 1976. "On ellipsis: Construct-state nouns in isolation." In:Ben-Zion FishIer & Raphael Nir, ads. Chaim Rabin Jubilee Volume.Jerusalem: Council on the Teaching of Hebrew. [Hebrew].

Schwarzwald, Ora R. 1981. Grammar 11M Reality in the Hebrew Varb. Ramat­GaIl: Bar-Ilan University. [Hebrew].

Schwarzwald, Ora. 1982. "�quency and regularity in language." IyyunimBeOhinuch 35, 163-174. [Hebrew].

Sivan, Reuben. 1980. The Revival of the Hebrew Language. Jerusalem:Rubinstein.

Smit.h, Karl. 1982. The use of compound adjectives in Hebrew, English, andNorwegian. Tel-Aviv University ma.

Werner, Fritz. 1982. "wortbildungstypen im Hebraischen". Folia Linguistica16, 263-295.

Werner, Fritz. 1983. Die Wortbildung der hebrai8chen Adjektive. Wiesbaden:Harr8S50witz.